Separation of Olefin/Paraffin Mixtures With Carrier Facilitated Membranes

Separation of Olefin/Paraffin Mixtures with Carrier
Facilitated Membranes
Contract No. DE-FC36-04GO14151
November 1, 2003 – October 31, 2006
Final report prepared by

Membrane Technology and Research, Inc.
1360 Willow Road, Suite 103
Menlo Park, CA 94025
for
U.S. Department of Energy

Golden Field Office
1617 Cole Blvd.
Golden, CO 80401
March 12, 2007
Contributors:
T. Merkel (PI)
R. Blanc
J. Zeid
A. Suwarlim
B. Firat
H. Wijmans
M. Asaro (SRI)
M. Greene (Lummus)
DE-FC36-04GO14151
Final Report
Covering Period: November 1, 2003 – October 31, 2006
Date of Report: March 14, 2007
Award Number: DE-FC36-04GO14151
Project Title: Separation of Olefin/Paraffin Mixtures with Carrier Facilitated

Membranes
Project Period: 11/01/2003 to 10/31/2006
Recipient Organization: Membrane Technology and Research, Inc.

1360 Willow Road, Suite 103
Menlo Park, CA 94025
Partners: SRI International ABB Lummus Global

333 Ravenswood Ave. 1515 Broad St.
Menlo Park, CA 94025 Bloomfield, NJ 07003
Technical Contact: Tim Merkel

650-328-2228 ext.134
tcmerkel@mtrinc.com
Business Contact: Elizabeth Weiss

650-328-2228 ext.142
egweiss@mtrinc.com
DOE Project Officer: William Prymak

Golden Field Office
1617 Cole Blvd.
Golden, CO 80401
303-275-4931
Bill.prymak@go.doe.gov
DOE Contract Specialist: Jean Siekerka

303-275-4924
Jean.siekerka@go.doe.gov
2 311 Final report

DE-FC36-04GO14151
TABLE OF CONTENTS
EXECUTIVE SUMMARY .............................................................................................................4

1. INTRODUCTION .................................................................................................................... 6
2. TECHNICAL BACKGROUND............................................................................................... 9
3. EXPERIMENTAL PROCEDURES....................................................................................... 12
3.1 Preparation of the Polymer Electrolyte Composite Membranes................................. 12
3.2 Membrane Stamp Permeation Measurements ............................................................ 14
4. INITIAL PERMEATION AND SORPTION PROPERTIES OF SILVER SALT
FACILITATED TRANSPORT MEMBRANES.................................................................... 15
4.1 Effect of AgBF4 Content on the Initial Pure and Mixed-Gas Properties of Pebax®
2533+AgBF4 Membranes ........................................................................................... 15
4.2 Effect of Multi-Component Feed Mixtures and Pressure on the Initial Permeation
Properties of Pebax® 2533+AgBF4 Membranes ......................................................... 20
4.3 The Impact of Salt Addition on Non-Facilitated Gas Permeances and Selectivities
of Pebax® 2533 Composite Membranes ..................................................................... 22
4.4 Olefin and Paraffin Sorption in Pebax® 2533+AgBF4 Membranes............................ 23
5. TIME-DEPENDENT PERMEATION IN SILVER SALT FACILITATED TRANSPORT
MEMBRANES ....................................................................................................................... 25
5.1 Background and Initial Findings................................................................................. 25
5.2 High Silver Salt Content Facilitated Transport Membranes are Not Stable Over
Time with Ideal Olefin/Paraffin Gas Mixtures ........................................................... 29
5.3 Understanding the Mechanism of Performance Decline in Silver Salt Facilitated
Transport Membranes – Olefin Conditioning............................................................. 32
5.4 Investigating the Reversibility of Olefin Conditioning .............................................. 36
5.5 FTIR Analysis of Olefin Conditioning ....................................................................... 40
5.6 Effect of Temperature, Polymer Plasticizers and Humidified Feed ........................... 41
5.7 Sorption Hysteresis ..................................................................................................... 47
5.8 Summary of Time-Dependent Membrane Studies ..................................................... 49
6. CARRIER POISONING AND REGENERATION OF SILVER SALT FACILIATED
TRANSPORT MEMBRANES............................................................................................... 49
6.1 Carrier Poisoning Studies ........................................................................................... 49
6.2 Stabilization and In-situ Regeneration of Reduced Silver Salt Membranes............... 54
7. NEW OLEFIN CARRIERS.................................................................................................... 59
7.1 Permeation Properties of Membranes Containing Alternate Olefin Carrier Salts...... 59
7.2 Permeation Properties of Membranes Containing Mixed Carrier Salts...................... 60
7.3 Development of Non-Silver Salt Olefin Carriers........................................................ 64
8. OLEFIN/PARAFFIN SEPARATION APPLICATIONS ...................................................... 71
8.1 Introduction................................................................................................................. 71
8.2 Olefin Metathesis ........................................................................................................ 72
9. CONCLUSIONS..................................................................................................................... 74
3 311 Final report

DE-FC36-04GO14151
EXECUTIVE SUMMARY
This document describes the results of a DOE-funded joint effort of Membrane Technology and
Research Inc. (MTR), SRI International (SRI), and ABB Lummus (ABB) to develop facilitated
transport membranes for olefin/paraffin separations. In the United States, ethylene and
propylene are produced in larger quantities than any other organic chemicals. Currently, olefins
and paraffins are separated by distillation—an extremely energy-intensive process because of the
low relative volatilities of olefins and paraffins. A new type of carrier facilitated membrane
discovered by MTR offers potential energy savings of 48 trillion Btu per year by the year 2020 if
this membrane technology could be successfully commercialized. As a step toward development
of these membranes, a detailed analysis of membrane time-dependent performance, including
performance after exposure to carrier poisoning species, was conducted during this project.
We discovered that silver salt-based facilitated transport membranes are not stable even in the
presence of ideal olefin/paraffin mixtures. This decline in membrane performance appears to be
caused by a previously unrecognized phenomenon that we have named olefin conditioning. As
the name implies, this mechanism of performance degradation becomes operative once a
membrane starts permeating olefins. Unfortunately, the very species targeted for separation by
the carrier membranes (olefins) are a source of membrane instability.
This project is the first study to identify olefin conditioning as a significant factor impacting the
performance of facilitated olefin transport membranes. To date, we have not identified an
effective strategy to mitigate the impact of olefin conditioning, other than running at low
temperatures or with low olefin feed pressures. In our opinion, this issue must be addressed
before further development of facilitated olefin transport membranes can proceed.
In addition to olefin conditioning, traditional carrier poisoning challenges must also be

overcome. Light, hydrogen, hydrogen sulfide, and acetylene exposure adversely affect
membrane performance through unwanted reaction with silver ions. Harsh poisoning tests with
these species showed useful membrane lifetimes of only one week. These tests demonstrate a
need to improve the stability of the olefin complexing agent to develop membranes with
lifetimes satisfactory for commercial application.
A successful effort to improve membrane coating solution stability resulted in the finding that
membrane performance loss could be reversed for all poisoning cases except hydrogen sulfide
exposure. A novel in-situ regeneration treatment restored aged and poisoned membranes to
nearly their initial performance values. This discovery offers the potential to extend membrane
lifetime through cyclic regeneration. We also found that certain mixed carriers exhibited greater
stability in reducing environments than that exhibited by silver salt alone. These results offer
promise that solutions to deal with carrier poisoning are possible.
The main achievement of this program was the progress made in gaining a more complete
understanding of the membrane stability challenges faced in the use of facilitated olefin transport
membranes. Our systematic study of facilitated olefin transport uncovered the full extent of the
stability challenge, including the first known identification of olefin conditioning and its impact
on membrane development.
4 311 Final report

DE-FC36-04GO14151
We believe that significant additional fundamental research is required before facilitated olefin
transport membranes are ready for industrial implementation. The best-case scenario for further
development of this technology would be identification of a novel carrier that is intrinsically
more stable than silver ions. If the stability problems could be largely circumvented by
development of a new carrier, it would provide a clear breakthrough toward finally recognizing
the potential of facilitated olefin transport.
5 311 Final report

DE-FC36-04GO14151
1. INTRODUCTION
Olefins are the most important building blocks of the petrochemical industry. In the United
States, ethylene and propylene are produced in larger quantities than any other organic chemical
– approximately 48 and 28 billion pounds annually, respectively. Most olefins are produced by
steam cracking of ethane, propane, naphtha, or gas oil. Production is highly energy intensive.
Current distillation processes for the separation of ethylene/ethane and propylene/propane
mixtures—which account for most of the energy consumption and much of the capital cost of the
steam cracker—consume an estimated 120 trillion Btu of energy per year in the United States.
The difficulty of the olefin/paraffin separation step stems from the low relative volatilities of the
components. As a result, very large distillation towers with 120-180 trays and high reflux ratios
are required. Even small improvements in these separations could result in significant energy and
cost savings to the petrochemical industry.[1-3]
In light of the importance of olefin/paraffin separations, and because of the costliness of

distillation, alternative technologies have been and continue to be the focus of industry study.
Two recent reviews by Eldridge[2, 3] describe the current status of research. Reactive absorption,
one of the more promising new techniques, uses silver or copper salts to extract the olefin from
bulk gas in an absorber/stripper process.[4] Union Carbide and BP/Stone and Webster have taken
this type of process to the pilot plant stage. Other processes that have been evaluated include
extractive distillation and molecular sieves.[5] Despite this work, however, a viable alternative to
conventional distillation has yet to be developed.
Membranes, an energy efficient, environmentally friendly technology, are potentially useful for
olefin/paraffin separations. Although polymeric membranes have been used successfully in
several gas separation applications, including nitrogen production from air and hydrogen
removal from refinery streams, the selectivities and gas fluxes of such membranes are inadequate
for separating olefins from saturated hydrocarbons. Under industrial operating conditions, the
best polymeric membranes offer olefin/paraffin selectivities of four to five. To replace or
supplement distillation for the separation of olefins and paraffins in steam crackers or propane
dehydrogenation plants, membranes with selectivities of 10 or more are necessary.
Facilitated transport membranes have long attracted research interest because of their potential to
perform separations more efficiently than simple, passive membranes. In a facilitated transport
process, passive diffusion down a concentration gradient is aided by the presence of a carrier
agent that selectively and reversibly binds with a compound targeted for separation, enhancing
its movement across a barrier. The complexation reaction of the carrier with the facilitated
species creates another transport route in addition to the typical solution-diffusion mechanism
that occurs in dense passive membranes. This second transport avenue is not available to non-
facilitated or inert species that are unable to interact with the carrier. Consequently, facilitated
transport membranes can exhibit selectivities well beyond the bounds that govern typical passive
membranes. For example, mixed-gas ethylene/ethane selectivities of more than 50 have been
demonstrated.[6, 7]
The spectacular separation performance of facilitated transport membranes has generated

considerable interest from academic and industrial groups. Over the past 40 years, hundreds of
6 311 Final report

DE-FC36-04GO14151
papers have been published describing membranes for facilitated transport of various gases. A
pioneering paper in this area was published by Scholander in 1960.[8] This study reported
facilitated oxygen transport in supported liquid membranes containing hemoglobin as the carrier
agent. Scholander found an eight-fold increase in oxygen flux for membranes containing
hemoglobin solution compared to those containing water only.[8] A few years later, Ward and
Robb reported facilitated carbon dioxide transport across immobilized liquid films containing an
aqueous bicarbonate carrier solution.[9] These membranes exhibited a selectivity of 4,100 for
carbon dioxide over oxygen. Like Scholander’s membrane, the carrier solution Ward and Robb
used was immobilized or supported within the pores of a microporous membrane. This same
concept was extended to facilitated olefin transport membranes by Steigelmann and Hughes at
Standard Oil, who used an aqueous silver (I) ion carrier contained within the pores of a cellulose
acetate film or fiber.[10, 11] These authors continued working on their facilitated olefin transport
membranes for over a decade, and a serious effort was made to commercialize the technology.[12]
The primary drawback to these supported liquid membranes is that over time the aqueous carrier
solution will evaporate or otherwise be pushed out of the membrane pores.[13] When this
happens, gases can pass non-selectively through the open pores. To improve stability and allow
reasonable lifetimes, the membranes have to be made relatively thick, which has the undesirable
effect of lowering transmembrane gas flux. Other means of improving the stability of supported
liquid facilitated transport membranes have been explored, including adding dense sealing
layers,[14] an absorbent liquid sweep,[12] or less volatile carrier solvents.[15] Nevertheless, the
inherent instability of supported liquid membranes is a serious deficiency that limits their
commercial potential.
In 1980, LeBlanc and coworkers[16] published an important paper describing facilitated olefin
transport in a dense ion-exchange membrane saturated with water. Silver was used as the
counter ion in a sulfonated poly(dimethylphenylene oxide) membrane. An advantage of this
membrane configuration is that the carrier species cannot be easily forced out of the membrane
because it is held in place within the polymer by electrostatic forces. In the following years, a
number of groups studied this type of facilitated transport membrane.[17-20] Although their
stability is better than the supported liquid membrane configuration, the ion-exchange
membranes require humidification to achieve facilitated transport. Without water, the silver
cation is held too tightly by the polymer exchange site (frequently SO3-), and is unable to interact
with olefins. Moreover, the ion-exchange polymers used in these membranes tend to be
expensive.
Ho overcame some of this cost hurdle by showing that relatively cheap hydrophilic polymers,
such as polyvinyl alcohol, containing silver salts could facilitate olefin transport when water
swollen.[21, 22] Around this same time, Peinemann reported high olefin/paraffin selectivities for
humidified membranes with a selective layer composed of a simple physical blend of Nafion®
and AgBF4.[23] Soon afterwards, Pinnau and Toy showed that hydrophilic polymers with
functional groups capable of solvating silver ions, such as polyethers, could facilitate olefin
transport even without humidification.[6, 7] Kang and co-workers examined facilitated olefin
transport in these dry solid polymer electrolyte membranes in detail.[24, 25] Recently, they have
shown that even hydrophobic polymers, such as polydimethylsiloxane, containing dispersed
silver salts can exhibit facilitated olefin transport.[26]
7 311 Final report

DE-FC36-04GO14151
Despite the progress outlined above, a number of daunting challenges remain before facilitated
olefin transport membranes can be used in industry. The most well-known of these problems is
the lack of carrier stability. In addition to complexing with olefins, known carriers also tend to
react with other species causing undesirable carrier deactivation or poisoning over time. For
example, the silver (I) ion – the most commonly-used carrier for facilitated olefin transport – can
be reduced to inactive silver metal by exposure to light or reducing gases such as hydrogen.[27]
Silver ions can also react with contaminants such as hydrogen sulfide or acetylene to form non-
facilitating, and in the case of silver acetylide, potentially dangerous, compounds.[4] This carrier
poisoning problem is often ignored or only briefly alluded to in the multitude of facilitated
transport membrane papers published each year.
The few detailed descriptions of olefin carrier poisoning are mostly found in industrial studies of
absorption systems. The idea of using silver (I) or other metal ions that can complex with olefins
to separate these species from mixtures dates to the early 20th century.[28-30] In 1951, Dewar
explained the metal-olefin π-bonding mechanism responsible for these complexes. Olefin
separation processes using aqueous silver salts as the absorbing solution in a packed column
were developed by Farbwerke Hoechst[27] in the 1950s and 1960s and Union Carbide[4] in the
1970s and 1980s. At both companies, the issue of carrier poisoning was addressed. To mitigate
the reduction of silver ions caused by reducing gases, hydrogen peroxide was added to the
absorbing solution in both company’s processes.[27, 31] Keller et al. at Union Carbide discovered
a method of managing silver acetylide formation through oxidation with silver permanganate.[31]
Keller suggested that the only solution to silver ion poisoning by sulfur compounds was to use
pretreatment to reduce these species to very low concentrations.[4] Although thorough
engineering solutions were worked out and large demonstrations of the technology conducted,
neither the Hoechst nor Union Carbide process caught on commercially, presumably due to
stability difficulties and cost.
The most detailed discussion of carrier poisoning in a facilitated olefin transport membrane
process comes from the work of Hughes et al. at Standard Oil.[12] This group also used hydrogen
peroxide to stabilize silver ions contained in their supported liquid membrane.[32] The
technology was field tested on the vent stream of a polypropylene plant using commercial-scale
hollow fiber modules impregnated with silver nitrate (AgNO3) solution. Even though a liquid
hexane sweep was used to mitigate solvent leak from the fibers, membrane deactivation due to
water loss was shown to adversely affect long-term performance. Lack of tolerance to poisons
such as H2S and acetylenes was also acknowledged as a process shortcoming.[12] Later, Davis et
al. at British Petroleum tried a similar supported liquid membrane contactor approach using a
recirculating AgNO3 solution.[33] This process also failed to achieve commercial success due to
cost and carrier stability problems.
A few groups have explored different olefin complexing agents in the hope of identifying a more
stable carrier for facilitated transport membranes. The most commonly studied carriers other
than silver are copper salts. Copper (I) ions form π-bond complexes with olefins in the same
manner that silver does. Unfortunately, copper carriers also suffer from stability problems.[3]
Alternate carriers that function via a different complexing mechanism have been examined as
well. In a patent assigned to the University of Colorado, DuBois and coworkers describe
molybdenum sulfide dimers that reversibly complex olefins and are unaffected by H2S.[34]
8 311 Final report

DE-FC36-04GO14151
Unfortunately, these materials are difficult to prepare and bind tightly with alkynes. Wilson et
al. report reversible olefin complexation for a palladium-containing cubane compound.[35] This
material is stable in the presence of hydrogen and acetylenes, but degrades upon exposure to H2S
or oxygen.
Membrane Technology and Research, Inc. (MTR) has continued to investigate the solid polymer
electrolyte facilitated transport membranes first described by Pinnau and Toy.[6, 7] These
composite membranes, consisting of a thin selective layer of silver salt dissolved in polyethers
and coated on a microporous support, can give useful permeances and selectivities without
humidification. However, as with previous silver salt membranes, they are subject to limited
membrane lifetimes because of carrier poisoning.
In this report, we describe our efforts to understand carrier poisoning and other membrane
stability issues as a first step in the scale-up and development of these facilitated transport
membranes for commercial use. A novel regeneration process capable of restoring the initial
properties of reduced facilitated transport membranes will be discussed. This discovery offers
the potential to periodically regenerate aged or poisoned membranes in a manner analogous to
backflushing of fouled filtration membranes. This encouraging finding is tempered by the
discovery of a previously unrecognized aging mechanism not related to carrier poisoning. The
causes of this time-dependent behavior, which is a serious barrier to further development of these
membranes, will be addressed in Section 5 of this report.
2. TECHNICAL BACKGROUND
The reversible reaction between transition metal ions and olefins has been known for more than a
century.[36] Metal salts, such as those of silver, form electron donor/acceptor complexes with
olefins through interaction of olefin π-orbitals with σ- and π-orbitals of the metal. Because they
react reversibly with olefins and their cost is relatively low, silver and copper salts have received
the most attention as potential olefin/paraffin separating agents.[3]
Previous work at MTR has identified silver tetrafluoroborate (AgBF4) as an effective olefin
complexing agent when dispersed in a polymer electrolyte. The mechanism by which facilitated
olefin separation takes place in a membrane is illustrated in Figure 1 for an ethylene/ethane
(C2H4/C2H6) mixture. On the high-pressure side of the membrane, both ethylene and ethane are
sorbed into the membrane material. Because ethylene forms a complex with the silver-ion
carrier, its solubility in the membrane is significantly higher than that of ethane. Desorption of
ethylene on the low-pressure side of the membrane leads to dissociation of the ethylene/silver
complex. Because ethane cannot form a complex with silver, it cannot access this facilitated
transport mechanism. At the same time that ethylene is transported selectively by silver
complexation, both ethylene and ethane are permeating by ordinary solution-diffusion. This
regular molecular permeation is relatively slow in comparison to that resulting from silver/olefin
complexation. As a result, membrane olefin/paraffin selectivities are very high.
9 311 Final report

DE-FC36-04GO14151
Figure 1. Schematic representation of a facilitated transport membrane containing silver

ions that react reversibly with olefins (ethylene, in this case).
While silver ions are known to form complexes with olefins, the exact mechanism by which the
ions transport olefin molecules across a dense membrane has been the subject of debate.
Previous findings suggest that the olefin/silver complex is relatively immobile in the
membrane.[6, 37, 38] Upon sorbing into the membrane, olefin molecules are believed to hop from
silver ion to silver ion, thereby traversing the film. The strongest evidence for this mechanism
comes from silver salt loading studies in which a percolation threshold is observed. Below a
critical loading, the space between silver ions is apparently too great for olefin molecules to hop
to a neighboring site, and no facilitated olefin transport occurs.
This result is illustrated in Figure 2, which shows the mixed-gas ethylene/ethane selectivity of a
polyether-polyamide block copolymer (Pebax® 2533) membrane that contains various amounts
of AgBF4. As the silver concentration in the polymer increases from 0 to 30 vol%, the
ethylene/ethane selectivity is close to 1 and nearly constant. At silver loadings of greater than 30
vol%, the olefin/paraffin selectivity increases dramatically. This happens to be the point at
which geometric percolation is reached for a simple dispersion of spheres within a medium. One
can imagine that a situation similar to that shown in the inset of Figure 2 is occurring. Below
percolation (<30 vol%), silver particles are separated by spaces filled with polymer. Permeating
olefin molecules complex with the silver ions, but then must diffuse through the polymer matrix
to reach the next silver particle. The diffusion step is relatively slow. In contrast, above
percolation (>30 vol%), the silver particles are in close communication, so olefin molecules can
hop from one silver ion to the next. This pathway is apparently much faster than diffusion
through the polymer phase, and consequently, olefin/paraffin selectivities are high above the
percolation threshold. For this reason, high loadings (typically around 80 wt% or 40 vol% of
silver salt) are used in our solid polymer electrolyte membranes.
10 311 Final report

DE-FC36-04GO14151
30
25
20
Mixed-gas
C2H4/C2H6
selectivity 15
geometric
percolation
10 threshold
0
0 10 20 30 40 50 60 70
AgBF4 concentration (vol%)
Figure 2. Mixed-gas ethylene/ethane selectivity in Pebax® 2533 as a function of AgBF4

content. Feed: 70 mol% ethylene/30 mol% ethane; Pressure: 50 psig;
Temperature: 22˚C.
The fact that membrane permeation properties appear to depend on the distribution of silver ions
within the matrix is important. As will be shown in Section 5, time-dependent changes in
membrane structure can significantly impact membrane performance.
Another way that membrane performance can be affected over time is through the presence of
carrier poisoning species. In addition to interacting with olefins, silver ions react with hydrogen,
acetylenes, hydrogen sulfide and other sulfur species.[3] These side reactions can deactivate the
silver carrier such that it is no longer able to complex olefins. As a result, membrane
olefin/paraffin selectivity is progressively lost with exposure time. The main side reactions that
are believed to occur are shown below; X is the anionic component of the silver salt (typically,
BF4-).
2Ag + + H 2 ⎯⎯→ 2Ag 0 + 2H + Hydrogen reduction

C 2 H 2 + 2AgX ⎯
⎯→ Ag 2 C 2 + 2HX Silver acetylide formation
H 2 S + 2AgX ⎯⎯→ Ag 2 S + 2HX Silver sulfide formation
In addition, silver salts can be reduced by light:
Ag + + e − ⎯⎯→
uv
Ag 0 Photo-reduction
11 311 Final report

DE-FC36-04GO14151
Despite the general understanding that these undesirable reactions can occur, there is very little
quantitative information available on the impact of carrier poisons on membrane performance.
This type of knowledge is important to gauge membrane lifetimes. In this report, we describe
the effect of light and carrier poisons on facilitated transport membrane performance. Methods
of mitigating the effect of these contaminants are also discussed.
3. EXPERIMENTAL PROCEDURES
3.1 Preparation of the Polymer Electrolyte Composite Membranes
The membranes used in this project were composite structures typically consisting of a thin
selective polymer-electrolyte/silver-salt layer sandwiched between a microporous support layer
and a protective top coat. A schematic drawing and an electron micrograph of a solid polymer
electrolyte facilitated transport membrane are shown in Figure 3. Mechanical strength is
provided by the poly(vinylidene fluoride) (PVDF) microporous support membrane. This support
is coated with a thin (~2 μm thick) selective layer composed of a polymer electrolyte (often
Pebax® 2533, a polyether-polyamide block copolymer composed of 80 wt% tetramethylene
oxide and 20 wt% nylon 12, and available from Arkema, Inc.) doped with a silver salt (AgBF4).
Finally, an extremely thin protective coating is applied on top of the selective Pebax®
2533+AgBF4 layer to seal defects and protect the membrane from damage during handling.
Figure 3. Schematic drawing and electron micrograph of a solid-state facilitated transport

composite membrane.
As shown in the electron micrograph, the inert PVDF support is quite porous and imparts
mechanical strength without providing significant resistance to mass transport. The matrix
polymer generally used for the selective layer, Pebax® 2533, was selected in a previous project
because it coated easily and appeared to yield the highest steady-state olefin permeances of the
various polyether-based materials examined. Other polymers examined have included
poly(ethylene oxide) (PEO), poly(vinyl pyrrolidone) (PVP), and poly(1-trimethylsilyl-1-
propyne) (PTMSP). Previously, we discovered that the Pebax® 2533+AgBF4 selective layer is
extremely hygroscopic causing the surface of these membranes to be very tacky. This surface
tackiness makes it difficult to handle Pebax® 2533+AgBF4 membranes. To change the surface
12 311 Final report

DE-FC36-04GO14151
properties, a hydrophobic protective layer was added to the composite membrane. This top layer
successfully altered the surface properties, allowing the membranes to be handled and rolled into
modules.
The microporous PVDF support was prepared with the casting machine shown in Figure 4. The
casting solution, which contains the polymer, a water-miscible solvent, and an additive, is
doctored onto a moving belt of polyester fabric. The belt is then collected on a take-up roll, after
which the membrane is washed overnight to remove any remaining solvent and, finally, dried in
an oven to form the support film.
Figure 4. Schematic diagram of the MTR casting machine.
The PVDF support layer was then coated with a solution of the polyelectrolyte and silver salt.
Acetonitrile and aliphatic alcohols are solvents for both Pebax® 2533 and AgBF4. Hence, thin-
film polymer electrolyte composite membranes were easily formed using a conventional dip-
coating process with the equipment shown schematically in Figure 5. The support membrane
passes from the feed roll through the dip-coating station, through a drying oven, and is finally
wound up on a product roll.
Figure 5. Schematic diagram of MTR’s thin-film coating machine.
Membrane thickness can be varied by adjusting the concentration of the coating solution and the
coating conditions. Normally, the thickness of the selective layer on a composite membrane is
13 311 Final report

DE-FC36-04GO14151
determined by comparing the flux of a composite membrane stamp with that of a homogeneous
film of known thickness. Because of the complex nature of the polymer electrolyte facilitated
transport membrane, it is somewhat difficult to determine thickness in this manner. Instead, we
used electron micrograph images, such as that shown in Figure 3, to estimate the thickness of the
selective layer. The selective layer coated from a 2 wt% Pebax® 2533+AgBF4 in ethanol
solution was ~2 μm thick.
3.2 Membrane Stamp Permeation Measurements
Figure 6 shows the system used to test Pebax® 2533+AgBF4 composite membrane stamps (area:
12.6 cm2) with gas mixtures. Each stamp was tested with a 50/50 ethylene/ethane mixture at a
feed pressure of 50 psig and a permeate pressure of 0 psig at 23°C unless otherwise noted. The
compositions of the feed, residue and permeate gases were analyzed with an on-line gas
chromatograph equipped with a thermal conductivity detector (TCD). Typically, the system is
operated at 1% stage cut (where stage cut equals the permeate flow rate divided by the feed flow
rate), so that the compositions of the feed and residue streams are essentially equivalent.
Because the permeation properties of silver salt facilitated transport membranes have been found
to depend on membrane preparation history, the method of measurement, and time, it is
important to compare results for stamps under similar conditions with similar histories. After
coating, each sample was dried in an oven at 50°C for one hour. The sample was then placed in
the test cell and operated at 1% stage cut for 30 minutes before a measurement was taken. The
value obtained at this time was taken as the initial or time zero reading.
Figure 6. Flow diagram of the mixed-gas permeation test system.
Early in this program, we realized that it would be necessary to test a large number of membrane
stamps for long periods of time in aging experiments. Consequently, a custom re-circulating,
multi-cell permeation test system was built for this project. This system is described in Section 5.
14 311 Final report

DE-FC36-04GO14151
4. INITIAL PERMEATION AND SORPTION PROPERTIES OF SILVER SALT

FACILITATED TRANSPORT MEMBRANES
4.1 Effect of AgBF4 Content on the Initial Pure and Mixed-Gas Properties of Pebax®
2533+AgBF4 Membranes
Previous work at MTR has shown that the addition of AgBF4 (silver tetrafluoroborate) to Pebax®
2533 (a block copolymer of nylon 12 and tetramethylene oxide) dramatically increases
olefin/paraffin selectivities in membranes fabricated from these materials. For example,
membranes made from Pebax® 2533 without AgBF4 have virtually no selectivity for ethylene
over ethane. In contrast, pure-gas ethylene/ethane selectivities of >100 and mixed-gas C2
selectivities of >30 are commonly observed in Pebax® membranes containing 80 wt% AgBF4.
The high selectivity of Pebax®+AgBF4 membranes results from the fact that AgBF4 addition
reduces paraffin permeance and, at high enough loadings, increases olefin permeance through
membranes. These transport effects are caused by the presence of dissolved Ag+ ions in the
membrane that selectively and reversibly complex olefin molecules. In doing so, the Ag+ ions
enhance olefin transport rates, while increasing the cohesive energy density of the matrix and
reducing the permeance of non-facilitated molecules (paraffins and other ‘inert’ species).
Despite this understanding, at the inception of this project there was very little quantitative
information on the effect of AgBF4 content on the transport rates of common gases through
facilitated transport membranes. Consequently, our first efforts were directed at developing a
basic set of performance data to provide this information.
Figure 7 presents the effect of AgBF4 content in Pebax® 2533 composite membranes on the pure-
gas permeances of hydrogen (H2), oxygen (O2), nitrogen (N2), methane (CH4), and carbon
dioxide (CO2). These data are also summarized in Table 1. As expected, the data confirm that
increasing amounts of AgBF4 reduce the permeances for all of these non-facilitated gases. The
largest decrease in permeance with increasing AgBF4 content occurs for methane, the largest
molecule in this group. This result indicates that AgBF4 increases the size-sieving nature of
Pebax® 2533, which is consistent with the idea that Ag+ ions increase the cohesive energy
density of the matrix. It is believed that Ag+ ions can form weak complexes with polar ether
groups in the polymer backbone. These complexes function as transient cross-links that restrict
polymer chain mobility. Consequently, the diffusion coefficients of permeating molecules are
reduced, with larger species being affected to a greater extent, leading to a more size-selective
membrane.
Table 1. Effect of AgBF4 Content on Pure-Gas Permeances of Inert Gases in Pebax® 2533
Membranes
AgBF4 Content Pure-Gas Permeances (gpu)

(wt%) H2 N2 O2 CH4 CO2
0 12 1.4 4.2 7.5 5.0
20 3.9 0.3 0.9 0.6 1.8
40 4.0 0.3 0.7 0.3 1.1
67 2.0 0.1 0.3 0.1 —
80 2.0 0.04 0.08 0.02 0.4
15 311 Final report

DE-FC36-04GO14151
50
Pebax® 2533
10
H2
Pure-gas
1
permeance
(gpu)
CO2
0.1 O2
N2
CH4
0.01
0 20 40 60 80 100
AgBF4 content (wt%)
Figure 7. Pure-gas permeances in Pebax® 2533+AgBF4 composite membranes as a function

of AgBF4 content in the dried selective layer. All measurements were taken at
room temperature with 100 psig feed pressure and 0 psig permeate pressure. All
coating solutions contained a small amount (<1 wt%) of H2O2/HBF4 stabilizer.
Consistent with this interpretation, the permeances of the larger paraffins, ethane and propane,
decrease at an even faster rate than that of methane as AgBF4 content increases, as demonstrated
in Figure 8. For example, propane (C3H8) permeance decreases by two orders of magnitude as
AgBF4 content increases from 0 to 40 wt% in Pebax® 2533, and is too low to be measured at
higher silver salt concentrations. Also shown in Figure 8 is the effect of AgBF4 content on
ethylene and propylene (C3H6) pure-gas permeances. Like the non-facilitated gases, at low silver
loadings the permeances of the olefins decrease. However, at higher silver loadings, this trend
reverses, and thereafter olefin permeance increases substantially with increasing silver salt
content. The increase in olefin permeance is due to the reversible reaction of olefins with silver
ions. This facilitated transport mechanism represents an additional permeation route accessible
to olefins but not to paraffins or other species that cannot complex with silver ions. The impact
of AgBF4 content on ethylene and propylene permeance can be explained in terms of the
competing effects of increased matrix energy density (which reduces permeance) and facilitated
transport (which increases permeance). Above a percolation threshold amount of AgBF4,
transport by complexing with silver ions dominates – perhaps reflecting the need for a minimum
density of Ag+ ions in the membrane to allow site-to-site hopping of olefin molecules.
Interestingly, the threshold appears to occur at different silver loadings for different olefin
molecules.
16 311 Final report

DE-FC36-04GO14151
500
Pebax® 2533
100
C2H4
C3H6
Pure-gas
10
permeance
(gpu)
1
C3H8
C2H6
0.1
0 20 40 60 80 100
AgBF4 content (wt%)
Figure 8. Pure-gas olefin and paraffin permeances in Pebax® 2533+AgBF4 composite

membranes as a function of AgBF4 content in the dried selective layer. All
measurements were taken at room temperature with 50 psig feed pressure and 0
psig permeate pressure All coating solutions contained a small amount (<1 wt%)
of H2O2/HBF4 stabilizer.
Figure 9 compares the effect of AgBF4 loading on the pure- and mixed-gas permeances of ethane
and ethylene. For the facilitated gas, ethylene, the permeation behavior in pure- and mixed-gas
experiments is very similar. In both cases, there is a percolation threshold around 30 wt%
AgBF4, beyond which permeances increase significantly with increasing silver salt content. For
the non-facilitated gas, ethane, there is more difference between the pure and mixed-gas results.
In both cases, ethane permeance decreases monotonically with increasing AgBF4 loading;
however, the pure-gas permeance decreases significantly more than the mixed-gas permeance.
For instance, at AgBF4 loadings of greater than 60 wt%, pure-gas ethane permeance is more than
an order of magnitude lower than mixed-gas ethane permeance.
17 311 Final report

DE-FC36-04GO14151
100 100
(a) ethane (b) ethylene
pure-gas
Gas permeance (gpu)

Gas permeance (gpu)
10
mixed-gas
10
mixed-gas
1
pure-gas
0.1 1
0 20 40 60 80 100 0 20 40 60 80 100
AgBF4 content (wt%) AgBF4 content (wt%)
Figure 9. Pure- and mixed-gas (a) ethane and (b) ethylene permeances in Pebax® 2533+
AgBF4 composite membranes as a function of AgBF4 content in the dried
selective layer. Pure-gas data collected at a feed pressure of 50 psig and 0 psig
permeate pressure. Mixed-gas data for a 65/35 C2H4/C2H6 mixture operating at
50 psig feed, 0 psig permeate and <1% stagecut.
The higher ethane permeance in mixture experiments is caused by swelling of the polymer
matrix resulting from extensive sorption of ethylene. This behavior is often seen in facilitated
membrane systems, and results in lower mixed-gas olefin/paraffin selectivities as compared to
ideal values (calculated from pure-gas experiments). Figure 10 illustrates this point, showing
that mixed-gas ethylene/ethane selectivities are lower than the corresponding pure-gas
selectivities for all AgBF4 loadings, and this difference increases with increasing silver salt
concentration.
18 311 Final report

DE-FC36-04GO14151
1000
Pebax® 2533
23°C
pure-gas
100
C2H4/C2H6
selectivity
mixed-gas
10
1
0 20 40 60 80 100
AgBF4 content (wt%)
Figure 10. Pure- and mixed-gas ethylene/ethane selectivities in Pebax® 2533+AgBF4

composite membranes as a function of AgBF4 content in the dried selective layer.
Pure-gas data were collected at a feed pressure of 50 psig and 0 psig permeate
pressure. Mixed-gas data are for a 65/35 C2H4/C2H6 mixture operating at 50 psig
feed, 0 psig permeate and <1% stagecut.
The Figure 10 results, showing an increase in mixed-gas ethylene/ethane selectivities with

increasing AgBF4 content, even at low silver loadings, differ from our previous results which
showed an increase in selectivity only at high silver salt loadings. A comparison of our new data
obtained in this project (data set R) with older results (data set A) from a previous NSF project is
shown in Figure 11.
The reasons for the performance differences are related to membrane treatment history and
measurement time. Olefin permeation rates and olefin/paraffin selectivities in silver salt
membranes are time dependent. In general, olefin permeances and olefin/paraffin selectivities
decrease over time. This behavior is related to a number of factors including changes in the
amount of solvent or atmospheric water present in the membranes, reduction of silver carrier
ions, and olefin-induced structural changes that occur in the membranes over time. These issues
will be discussed in more detail in Section 5. Although the treatment history and measurement
time for our previous results shown in Figure 11 (data set A) are unknown, the lower selectivities
are consistent with aged membranes.
To compare membrane properties throughout this report, unless otherwise stated, olefin
permeances and olefin/paraffin selectivities represent initial or as-prepared values (measured
19 311 Final report

DE-FC36-04GO14151
after drying freshly-coated membranes for 1 hour at 50°C, then exposing the membranes to the
standard feed mixture for 30 minutes at <1% stagecut). Time-dependent permeation properties
will be clearly identified as such.
100
Pebax® 2533
23°C
Data Set R:
Feed= 65%/35%
C2H4/C2H6 at 50 psig
C2H4/C2H6
selectivity 10
Data set A:
Feed = 70%/30%
C2H4/C2H6 at 50 psig
1
0 20 40 60 80 100
AgBF4 content (wt%)
Figure 11. Comparison of data sets for mixed-gas ethylene/ethane selectivities in Pebax®
2533+AgBF4 composite membranes as a function of AgBF4 content. All
experiments were conducted at 0 psig permeate and <1% stagecut. Data set R
measurements were taken after 1 hour oven drying followed by 30 minutes of
mixed-gas permeation. Sample preparation history and measurement time are
unknown for data set A.
4.2 Effect of Multi-Component Feed Mixtures and Pressure on the Initial Permeation
Properties of Pebax® 2533+AgBF4 Membranes
Pebax® 2533 composite membranes containing 80 wt% AgBF4 were evaluated with binary and
four-component gas mixtures at different feed pressures. Figure 12 illustrates the impact of feed
pressure on binary mixed-gas ethylene/ethane permeances and selectivity. Often, the selectivity
of facilitated transport membranes will decrease with increasing feed pressure because the carrier
species becomes saturated. This appears to be the case for our Pebax® 2533+80 wt% AgBF4
membranes. As feed pressure is increased, ethylene permeance decreases linearly, but ethane
permeance is roughly constant. As a result, ethylene/ethane selectivity also drops linearly, as
seen in Figure 12.
20 311 Final report

DE-FC36-04GO14151
100 70
(a) (b)
Mixed-gas C2H4/C2H6 selectivity

ethylene 60
Mixed-gas permeance (gpu)
10 50
40
ethane
1 30
20
0.1 10
40 60 80 100 120 140 40 60 80 100 120 140
Feed pressure (psig) Feed pressure (psig)
Figure 12. Mixed-gas ethane/ethylene (a) permeances and (b) selectivities in Pebax®
2533+80 wt% AgBF4 composite membranes as a function of feed pressure. All
experiments conducted at 0 psig permeate, room temperature, and <1% stagecut
with a 65/35 C2H4/C2H6 mixture.
Although carrier saturation effects are observed, Pebax® 2533+80 wt% AgBF4 membranes still
yield high selectivities at industrially relevant pressures. For example, a selectivity of about 30 is
obtained for an ethylene/ethane feed mixture at 100 psig. This differentiates these membranes
from other facilitated membrane systems where high selectivities are observed only at low, even
sub-atmospheric pressures.
Figure 13 presents the effect of feed pressure on the permeances and selectivities of Pebax®
2533+80 wt% AgBF4 membranes for a four-component mixture comprised of carbon dioxide,
methane, ethane, and ethylene. The mixed-gas permeances of these species follow the order:
C2H4 > CO2 > CH4 > C2H6. As expected, the facilitated gas—ethylene—is the most permeable
species, followed by the other gases in the order of increasing molecular size. In comparing the
data in Figures 7 and 13, it is evident that the mixed-gas permeances of the non-facilitated gases
(CO2, CH4, and C2H6) are roughly an order of magnitude higher than their pure-gas permeances.
This result is consistent with the idea that sorption of ethylene during co-permeation swells the
membrane, allowing non-facilitated gases to permeate faster. Also shown in Figure 13 is the
effect of feed pressure on ethylene/ethane selectivity. Similar to the data for a binary mixture
(Figure 12), mixed-gas ethylene/ethane selectivity decreases with increasing feed pressure for the
four-component mixture. The ethylene/ethane selectivities observed for the four-component
mixture are significantly lower than those observed in the binary case. The reasons for this
difference are not known.
21 311 Final report

DE-FC36-04GO14151
100 30
(a) (b)

25
C2H4
10 CO2 20
15
C2H6
1 CH4 10
0.1 0
40 60 80 100 120 140 160 40 60 80 100 120 140 160
Feed pressure (psig) Feed pressure (psig)
Figure 13. Four-component mixed-gas permeances and ethylene/ethane selectivities in

Pebax® 2533+80 wt% AgBF4 composite membranes as a function of feed
pressure. All experiments were conducted at 0 psig permeate, room temperature,
and <1% stagecut with a feed mixture composed of 33 mol% ethane / 32%
ethylene / 25% methane / 10% carbon dioxide.
4.3 The Impact of Salt Addition on Non-Facilitated Gas Permeances and Selectivities of
Pebax® 2533 Composite Membranes
To further examine the effect of salt addition on membrane properties, additional Pebax® 2533
composite membranes containing salts were prepared and tested with non-facilitated gases.
Pure-gas O2, N2, CH4, and CO2 permeances and selectivities for Pebax® 2533 membranes
containing AgBF4, AgNO3, and the lithium salts LiNO3 and LiCl are summarized in Table 2.
Compared to an unfilled Pebax® 2533 membrane, all of the salts except AgNO3 reduced gas
permeances. The extent of this permeance reduction varied depending on the salt. For example,
at 30 wt% salt loading, AgBF4 reduces CO2 permeance by 67%, LiNO3 reduces the permeance
by only 30% and AgNO3 appears to slightly increase permeance. These results are likely related
to the relative extent to which the salts are disassociated in Pebax® 2533. The large BF4- anion is
rather weakly held to Ag+ and thus most, if not all, of the silver ions are free to interact with the
polymer matrix and increase the cohesive energy density of the matrix. AgNO3, on the other
hand, forms a much tighter bond and is known to be only partially disassociated even in dilute
aqueous solution. Consequently, its silver ions are not as free to interact with ether groups in the
polymer as those present in AgBF4. LiNO3 has a lower dissociation energy than AgNO3, but
higher than that of AgBF4 and thus has an intermediate effect on permeance.
In addition to reducing gas permeances, incorporating salts into Pebax® 2533 membranes
generally increases size selectivity. For example, the CO2/CH4 and O2/N2 selectivities of Pebax®
2533 membranes containing 30 wt% LiNO3 are 15 and 3.5, respectively, compared to 9.2 and 2.7
22 311 Final report

DE-FC36-04GO14151
for the Pebax® membrane without salt. This result is consistent with the well-known trade-off
relationship between membrane permeance and ability of the matrix to discriminate between
molecules based on size differences. As salt is added to the membrane, the polymer cohesive
energy density increases, which in turn reduces diffusion coefficients and increases diffusion
selectivity. Consequently, the selectivity of the membrane for smaller molecules (O2 and CO2)
over larger molecules (N2 and CH4) increases with salt content.
Table 2. Effect of Salt Addition on Permeances and Selectivities of Pebax® 2533

Composite Membranes
Pure-gas Permeances Pure-gas Selectivity

Membrane type
N2 O2 CH4 CO2 O2/N2 CO2/CH4
Pebax® 2533 / PVDF 1.6 4.4 5.4 50 2.7 9.2

®
Pebax 2533 +
0.85 2.6 2.8 28 3.1 9.6
20 wt% LiNO3 / PVDF
Pebax® 2533 +
0.77 2.6 2.4 35 3.5 15
30 wt % LiNO3 / PVDF
Pebax® 2533 +
0.75 2.2 2.2 27 2.9 13
30 wt% LiCl / PVDF
Pebax® 2533 +
2.2 6.1 7.5 57 2.8 7.6
30 wt% AgNO3/ PVDF
Pebax® 2533 +
- - 1.7 17 - 10
30 wt% AgBF4 / PVDF
Pebax® 2533 +
- - 1.0 7.0 - 7.0
80 wt% AgBF4 / PVDF
4.4 Olefin and Paraffin Sorption in Pebax® 2533+AgBF4 Membranes
Figure 14 presents ethylene and ethane sorption isotherms in Pebax® 2533 films containing
varying amounts of AgBF4. As expected, the addition of AgBF4 dramatically increases the
amount of ethylene sorbed by Pebax® 2533. For example at 10 atm, Pebax® 2533 containing 70
wt% AgBF4 sorbs about 20 times as much ethylene as pure Pebax® 2533 (200 vs 10
cm3(STP)/cm3 polymer). This increase in ethylene sorption capacity is due to complexation of
the olefin by dispersed silver ions. Figure 14(b) shows that ethane solubility is depressed by the
addition of AgBF4. For this non-interacting gas, the dominant effect of AgBF4 addition is to
increase the polymer cohesive energy density. The resulting decrease in ethane sorption is an
example of the “salting-out” phenomenon. Interestingly, these sorption results suggest that even
small amounts of AgBF4 should yield membranes with much higher ethylene/ethane selectivities
than pure Pebax® 2533. The fact that this often does not happen (see Figure 2 for example)
suggests that ethylene may be held rather tightly by silver ions (yielding very low apparent
ethylene diffusion coefficients). We will return to this issue in section 5.
23 311 Final report

DE-FC36-04GO14151
250 15
(a) (b) 0 wt%
70 wt%
C2H6 concentration, cm3(STP)/cm3

200
50 wt% 10
150 10 wt%
30 wt%
100
5
30 wt%
50 10 wt%
0 wt% 50 wt%
0 0
0 5 10 15 0 5 10 15
C2H4 pressure, atm C2H6 pressure, atm
Figure 14. (a) Ethylene and (b) ethane sorption isotherms in Pebax® 2533 films containing
varying amounts of AgBF4 at 25°C.
The amount of ethylene sorbed by Pebax® 2533 membranes containing large amounts of silver
salt is extremely high. Figure 15(a) illustrates this fact by comparing ethylene sorption in
Pebax® 2533 containing 70 wt% AgBF4 with that in poly(1-trimethylsilyl-1-propyne) (PTMSP).
PTMSP is a high-free-volume glassy material with the highest sorption capacity of all known
polymers. Nevertheless, Pebax® 2533+70 wt% AgBF4 sorbs two to three times more ethylene
than PTMSP. Another way to view ethylene sorption in silver salt membranes is shown in
Figure 15(b). Here, ethylene sorption is expressed as the volume percent of liquid ethylene in
the polymer/salt blend. At ethylene pressures typically encountered on the feed side of a Pebax®
2533+AgBF4 composite membrane, the amount of ethylene in the selective layer exceeds 25
vol%. In this situation, the polyether/silver ion blend is practically swimming in a sea of
ethylene. This highly plasticized environment suggests that some mobility of polymer and ionic
species within the membrane is likely. Such mobility has implication for long-term membrane
stability especially under pressure cycling.
24 311 Final report

DE-FC36-04GO14151
250 40
(a) Pebax® 2533 + (b)
70 wt% AgBF4
35
200 70 wt% AgBF4
C2H4 concentration, vol%

30
150 25
20
100
PTMSP 15
10
50
5
Pebax ® 2533
0 0
0 5 10 15 0 5 10 15
C2H4 pressure (atm) C2H4 pressure (atm)
Figure 15. (a) Ethylene sorption isotherms in Pebax® 2533, Pebax® 2533 containing 70 wt%
AgBF4, and high free volume glassy PTMSP at 25°C. (b) Ethylene sorption in
Pebax® 2533 containing 70 wt% AgBF4 expressed as volume percent of sorbed
liquid ethylene.
5. TIME-DEPENDENT PERMEATION IN SILVER SALT FACILITATED

TRANSPORT MEMBRANES
5.1 Background and Initial Findings
The long-term instability of facilitated transport membranes has previously been recognized as
the key obstacle preventing commercial use of this technology. At the inception of this project,
the general belief was that deactivation of the silver ion carriers by reducing gases (H2, H2S,
C2H2) was the primary cause of membrane instability. The effect of these carrier poisons, and a
novel solution to mitigate their impact, are discussed in Section 6. In this section, previously
undocumented time-dependent membrane behavior not related to carrier poisoning is discussed.
Early in this program, we recognized that examining performance changes over long time frames
in an efficient manner required a re-circulating, multi-cell permeation system. Figure 16 shows a
picture of the system built for this purpose. After calibrating the new system, the first topic of
investigation was the effect of time and AgBF4 content on membrane selectivity with an ideal
ethylene/ethane mixture (no carrier poisons). Previously, two data sets were shown (Figure 11)
describing the impact of AgBF4 concentration on membrane selectivity that were seemingly
contradictory. One set of data (labeled set A) appeared to show a silver concentration
percolation threshold, below which the membranes had essentially no olefin/paraffin selectivity.
The second data set (labeled set R) showed a monotonic increase in olefin/paraffin selectivity as
AgBF4 concentration increased. It was suggested that this difference might be related to
25 311 Final report

DE-FC36-04GO14151
differences in membrane preparation or the measurement time. Both sets of data in Figure 11
were measured in a single-pass experiment where the feed gas is drawn directly from a gas
cylinder. The data set R results were collected after permeating gas at 1% stage cut for 30 min,
while the measurement time for set A was unknown.
Figure 16. Re-circulating, multi-cell membrane test system.
Figure 17 presents mixed-gas selectivity data for membranes containing different AgBF4
concentrations, measured over a period of five days on the re-circulating permeation system.
The initial data, collected a few hours after startup, appear very similar to data set R in Figure 11.
Over a period of several days, as the system runs continuously, mixed-gas selectivities decrease,
especially at AgBF4 contents below 60 wt%. After four days, the effect of silver content on
mixed-gas ethylene/ethane selectivity is very similar to that observed for data set A. For
example, the sample containing 40 wt% AgBF4, which originally had an ethylene/ethane
selectivity of 14, has a selectivity of 1.2 after four days of continuous operation. This is
consistent with the values measured in data set A. The membranes containing 80 wt% AgBF4
appear to be less affected by this reduction in selectivity over time. However, even at this high
silver concentration, performance does deteriorate over the course of the experiment.
The reasons for this performance decrease over time with an ideal gas mixture are discussed in
the remainder of this section. Initially, a number of possible explanations were proposed,
including
26 311 Final report

DE-FC36-04GO14151
• the presence of residual solvent or atmospheric water in the freshly-coated membranes,

• morphological changes in the polymer-salt blend upon olefin exposure, or
• slow carrier salt recrystallization during operation.
All of these mechanisms may play some role, although as will be shown, the evidence points
largely toward olefin-induced morphological changes in the polymer-salt blend. This
phenomenon not only results in a decrease in olefin/paraffin selectivity over time (even at 80
wt% AgBF4), but perhaps more importantly, it also causes a significant decrease in olefin
permeance, as discussed below.
100
Pebax® 2533
23°C
operating original
MTR data
Mixed-gas time
C2H4/C2H6
10
selectivity 0 days
1 day
2 days
3 days
4 days
1
0 20 40 60 80 100
AgBF4 content (wt%)
Figure 17. Effect of operating time and AgBF4 content on the mixed-gas ethylene/ethane
selectivity of Pebax® 2533 composite membranes.
Figure 18 presents the effect of operating time on ethylene permeance for membranes containing
various levels of AgBF4. Two different types of time-dependent behavior are evident:
• the 0 and 20 wt% AgBF4 samples exhibit stable ethylene permeances, and
• the higher silver salt content samples demonstrate decreasing permeances over time.
The initial (time zero) impact of silver salt on ethylene permeance can be rationalized as the
interplay between two competing effects. The addition of an ionic salt species increases the
cohesive energy density of the polymer matrix, which decreases diffusion coefficients. Increased
cohesive energy density is the dominant effect as AgBF4 content is increased from 0 to 20 wt%,
and consequently, ethylene permeance decreases. At higher silver salt contents, increased
27 311 Final report

DE-FC36-04GO14151
cohesive energy density (which reduces permeance) is offset by the increased carrier density that
allows site-to-site hopping of olefin molecules (which increases ethylene permeance). As a
result, the initial ethylene permeance exhibits the following trend with silver salt content:
80 wt% > 60 wt% > 40 wt% ≈ 0 wt% > 20 wt%.
Over time, the ethylene permeances of the high silver salt content samples (40, 60, and 80 wt %)
decrease substantially. It is this decrease in ethylene permeance that is the primary cause of the
selectivity decline illustrated in Figure 17. It is interesting to note that after four days of
operation, while the ethylene permeance has decreased in the 40, 60, and 80 wt% AgBF4
samples, their permeance relative to each other is nearly unchanged (i.e., 80 wt% > 60 wt% > 40
wt%). This result suggests the hopping mechanism of ethylene transport is still operative in the
aged membranes because the highest carrier density sample (80 wt %), where ethylene hopping
is easiest, retains the highest permeance throughout the test period.
100
0 wt%
10
Mixed-gas 20 wt%
ethylene
permeance
(gpu)
1 80 wt%
60 wt%
40 wt%
0.1
0 1 2 3 4 5
Operating time (days)
Figure 18. Mixed-gas ethylene permeance as a function of operating time for Pebax® 2533
membranes containing various concentrations of AgBF4.
Figure 19 presents the effect of operating time on ethane permeance for membranes containing
various amounts of AgBF4. Ethane does not form a complex with silver ions, and consequently,
the only impact of increasing silver salt content should be to increase membrane cohesive energy
density, which reduces permeance. This is essentially the case, as the initial ethane permeance
exhibits the following trend with silver salt content:
0 wt% > 20 wt% > 40 wt% ≈ 60 wt% ≈ 80 wt%.
28 311 Final report

DE-FC36-04GO14151
Similar to the ethylene permeance results, ethane permeances in the 0 and 20 wt% AgBF4
samples are relatively unchanged over the course of the experiment. However, the 40, 60, and
80 wt% AgBF4 samples exhibit an ethane permeance decline over time, similar to that observed
for ethylene. Because ethane cannot form complexes with silver ions, it is not able to access the
carrier hopping transport mechanism. The fact that the permeance of the non-facilitated gas
(ethane) decreases with time indicates that carrier deactivation is not the cause of the flux decline
observed in Figures 18 and 19. Rather, there is something else that affects both the facilitated
and non-facilitated species.
0 wt%
10
20 wt%
Mixed-gas 1
ethane
permeance
(gpu) 80 wt%
60 wt%
40 wt%
0.1
0.01
0 1 2 3 4 5
Operating time (days)
Figure 19. Mixed-gas ethane permeance as a function of operating time for Pebax® 2533
membrane containing various concentrations of AgBF4.
5.2 High Silver Salt Content Facilitated Transport Membranes are Not Stable Over
Time with Ideal Olefin/Paraffin Gas Mixtures
In light of the findings shown in Figures 17-19, additional long-term permeation tests of the
baseline membrane (Pebax® 2533+80 wt% AgBF4) were conducted. Figure 20 presents time-
dependent facilitated transport membrane results previously obtained at MTR. These data are
compared with new results (Figure 21) collected in this program with our recirculating
permeation system. Both sets of data are for 80 wt% AgBF4 in Pebax® 2533 composite
membranes using an ideal ethylene/ethane gas mixture. The results in Figure 20 show an initial
permeance and selectivity decline over a period of two days followed by stable behavior
thereafter. In contrast, the data of Figure 21 show a continuous decrease in permeance and
selectivity over time. This result is surprising and contrary to previous results for AgBF4-based
facilitated transport membranes reported here and in the literature.
29 311 Final report

DE-FC36-04GO14151
100 50
T=22°C (a) T=22°C (b)
Mixed-gas C 2H4/C2H6 selectivity

40
10
Ethylene 30
20
1
10
Ethane
0.1 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Exposure time (days) Exposure time (days)
Figure 20. Mixed-gas (a) ethylene and ethane permeances and (b) ethylene/ethane selectivity
for Pebax® 2533+80 wt% AgBF4 membranes as a function of time.
Measurements were conducted with a 70% ethylene / 30% ethane mixture at a
feed pressure of 50 psig. Data were collected in a previous NSF project.
100 50
(a) (b)
Mixed-gas permeance, GPU
40
10
Ethylene
30
20
Ethane
0.1
10
0.01 0
0 1 2 3 4 5 0 1 2 3 4 5
Permeation time (days) Permeation time (days)
Figure 21. Mixed-gas (a) ethylene and ethane permeances and (b) ethylene/ethane selectivity
for Pebax® 2533+80 wt% AgBF4 membranes as a function of time.
Measurements were conducted with a 50% ethylene / 50 % ethane mixture at a
feed pressure of 50 psig on the recirculating permeation test system.
30 311 Final report

DE-FC36-04GO14151
The primary difference between the new and old results is the method of measurement. In all
previous studies that we are aware of, mixed-gas permeance in silver salt facilitated transport
membranes has been measured using cylinder gas to supply pressure to the feed side of the
membrane. For these measurements, a high flow rate of gas must be passed across the feed side
of the membrane to prevent depletion of the highly-permeable olefin in the mixture. Typically,
the experiment is conducted at <1% stage cut, where >99% of the gas mixture fed to the
membrane is vented out the residue side of the permeation cell. In the very few studies of long-
term membrane performance, such as our data shown in Figure 20, the residue and permeate
valves are typically closed between measurements to conserve the gas mixture. In contrast, for
the recirculating permeation system, the residue and permeate gases are combined, compressed,
and recirculated to the feed side of the membrane continuously. With this system, the residue
and permeate gases can flow constantly without having to vent large amounts of the gas mixture.
The experimental difference in the new and old results obtained can be explained as follows: the
x-axis of the data in Figure 20 is actually a measure of feed gas exposure time, while for Figure
21 the x-axis is truly permeation time. It appears that this difference is significant for long-term
membrane performance, which will be important to successful industrial use of carrier facilitated
membranes. For actual commercial operation, the membrane must run continuously for long
permeation times.
To confirm our new results, we ran another long-term mixed-gas stamp test using a different
permeation system where test gases are also compressed and recirculated continuously. These
results, shown in Figure 22, are consistent with the data shown in Figure 21. The mixed-gas
permeances of olefin and paraffin (propylene and propane, respectively) decrease continuously
over time. Propylene permeance decreases faster, resulting in a steady decline in membrane
selectivity. After six days, the permeances are simply too low to measure accurately.
31 311 Final report

DE-FC36-04GO14151
1,000 25
(a) (b)
20
Mixed-gas C3 selectivity
100
15
Propylene
10
10
1
5
Propane
0.1 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Figure 22. Mixed-gas (a) propylene and propane permeances and (b) propylene/propane
selectivity for Pebax® 2533+80 wt% AgBF4 membranes as a function of time.
Measurements were conducted with a 50% propylene / 50% propane mixture at a
feed pressure of 50 psig using recirculated gas on MTR’s “beta” test system.
This new time-dependent membrane behavior when exposed to ideal olefin/paraffin mixtures
was surprising and is a significant concern with regard to the viability of using these membranes
industrially. To be a cost-competitive separation option, membrane modules must show
commercial lifetimes of at least one year. We had anticipated the technical challenge of carrier
deactivation by poisons such as hydrogen sulfide and acetylenes. This hurdle, and ways to
alleviate its effects are discussed in Section 6. However, performance decline over time with
ideal olefin/paraffin mixtures is a serious and separate issue that warranted further study.
5.3 Understanding the Mechanism of Performance Decline in Silver Salt Facilitated

Transport Membranes – Olefin Conditioning
Figure 23 presents pure-gas helium permeance as a function of permeation time in silver salt
membranes that utilize Pebax® 2533 or polyethylene oxide (PEO) as the matrix polymer. This
test was conducted to determine whether the permeance of a non-facilitated gas (one that does
not complex with Ag+ ions) through silver salt membranes is dependent on permeation time
when the membrane is not exposed to an olefin. We chose helium as the non-facilitated gas
because the pure-gas permeances of most other gases that do not complex with silver ions are too
low to accurately measure in membranes containing high concentrations of AgBF4. The data in
Figure 23 show that helium permeance is completely stable in Pebax® 2533 and PEO membranes
containing 80 wt% AgBF4, as well as in a control Pebax® membrane that does not contain silver
salt. This result is in contrast to mixed-gas permeance results for non-facilitated gases when an
olefin is present in the mixture. For example, in Figure 21(a), ethane, a non-facilitated gas,
32 311 Final report

DE-FC36-04GO14151
exhibits permeance decline over time when tested in a mixture with ethylene. A similar result is
shown in Figure 22(a) for mixed-gas propane permeance. These findings suggest that the
presence of an olefin, which can complex with silver ions, causes permeance decline in the
membrane and that it alters the membrane in a way that impacts non-facilitated gas permeances.
100
®
Pebax 2533
10
®
Helium 80 wt% AgBF4 in Pebax 2533
permeance
(gpu)
80 wt% AgBF4 in PEO
0.1
0 12 24 36 48
Permeation time (hr)
Figure 23. Pure-gas helium permeance as a function of permeation time in silver salt
facilitated transport membranes based on Pebax® and polyethylene oxide (PEO).
After 40 hours of permeating helium, the feed gas for the membranes described in Figure 23 was
switched to an ethylene/ethane mixture. Figure 24 shows the subsequent mixed-gas ethylene
permeance of these membranes as a function of permeation time. The control Pebax® 2533
membrane that does not contain AgBF4 exhibits stable ethylene permeance throughout the
experiment. In contrast, the Pebax® and PEO membranes containing 80 wt% AgBF4 show
significant permeance decline, similar to that illustrated in Figure 21(a). These results indicate
that regardless of the polymer matrix used (Pebax® or PEO), olefin-induced performance decline
appears to be common for silver salt facilitated transport membranes. Although not shown, the
permeance of the non-facilitated gas, ethane, also decreased continuously throughout these
mixture experiments.
At the end of the mixed-gas experiment described in Figure 24, the feed gas was switched back
to pure helium. Our goal in this experiment was to use helium as a probe molecule and see
whether we could detect changes in the membrane structure after it was exposed to ethylene. For
the control membrane that does not contain AgBF4, the helium permeances before and after
ethylene permeation were approximately equivalent. In contrast, the helium permeances for the
membranes containing 80 wt% AgBF4 were higher after ethylene permeation. For example, in a
33 311 Final report

DE-FC36-04GO14151
Pebax® 2533+AgBF4 membrane, helium permeance was 3 gpu before the ethylene mixed-gas
test and 9 gpu after this experiment. This result confirms that membranes containing silver salt
are altered by olefin permeation in a way that impacts gas permeances even after olefin
molecules are removed from the membrane. Subsequent attempts to cycle these membranes to
investigate the reversibility of the olefin conditioning effect were hindered by the formation of
membrane defects. We will return to the issue of reversibility later in this section.
100
®
Pebax 2533
10
80 wt% AgBF4
Mixed-gas in PEO
ethylene
permeance
(gpu)
80 wt% AgBF4
1
®
in Pebax 2533
0.1
0 24 48 72 96 120 144
Permeation time (h)
Figure 24. Mixed-gas ethylene permeance as a function of permeation time for two Pebax®
2533 membranes without AgBF4, two Pebax® 2533 membranes containing 80
wt% AgBF4, and a PEO membrane containing 80 wt% AgBF4.
In the preceding discussion, it was shown that permeation of non-facilitated gases through
Pebax®+AgBF4 membranes is completely stable over long periods of time. However, as soon as
the membrane begins permeating an olefin, the permeances of all species start to decline over
time. This result suggests that the olefin-Ag+ ion complexation reaction causes time-dependent
behavior. Because the permeances of non-facilitated gases are also affected when co-permeating
with an olefin, it appears that the olefin-Ag+ ion interaction causes membrane structural changes
that occur slowly over time.
Figure 25 presents mixed-gas ethylene permeance in Pebax® 2533+80 wt% AgBF4 membranes
as a function of permeation time for three different ethylene feed concentrations.
• When the feed gas is a 50/50 mixture of ethylene/ethane (yielding an ethylene partial
pressure of 32.5 psia), the ethylene permeance decreases relatively rapidly. With this
mixture, the ethylene permeance decreases by almost two orders of magnitude over 120
hours of permeation.
34 311 Final report

DE-FC36-04GO14151
• When the ethylene concentration in the feed gas is only 15% (and the ethylene feed
partial pressure is 9.8 psia), the rate of permeance decline is considerably slower. For
example, over a similar period of 120 hours, the ethylene permeance decreases about one
order of magnitude.
• Finally, if the ethylene concentration is reduced to 7% (an ethylene partial pressure of 4.5
psia), ethylene permeance is essentially stable during 280 hours of permeation.
These results are consistent with the idea that the olefin complexation reaction with Ag+ ions is
responsible for the time-dependent membrane performance. Previously, we have shown that
membranes containing less than 40 wt% AgBF4 exhibit stable permeation properties, although
selectivities are low. Unfortunately, membranes with a high silver salt content, which is
necessary for adequate olefin/paraffin selectivities, are not stable in the presence of olefins – the
very species the membranes are being used to separate – unless the olefin feed concentration (or
partial pressure) is low.
100
7% C2H4 feed
(pi = 4.5 psia)
Mixed-gas
ethylene 10
permeance
(gpu) 15% C2H4 feed
(pi = 9.8 psia)
50% C2H4 feed

(pi = 32.5 psia)
0.1
0 50 100 150 200 250 300
Figure 25. Mixed-gas ethylene permeance in Pebax® 2533 membranes containing 80 wt%
AgBF4 as a function of permeation time for three different feed mixtures. Tests
were conducted at room temperature with 50 psig feed pressure and 0 psig
permeate pressure.
Figure 26 shows the mixed-gas ethylene/ethane selectivities of Pebax® 2533+80 wt% AgBF4
membranes for the same feed conditions described in Figure 25. There is considerable scatter in
these data; nevertheless, the general trend appears to be similar to the ethylene permeance
results. As the ethylene concentration (or partial pressure) in the feed mixture increases, the rate
35 311 Final report

DE-FC36-04GO14151
of ethylene/ethane selectivity decline increases. Together, the data in Figures 25 and 26 provide
further evidence for an “olefin conditioning” effect that is a cause of the decline in facilitated
transport membrane performance. Because this conditioning is caused by olefins but affects the
permeation of non-facilitated species, it suggests that the silver ion-olefin complexation reaction
progressively alters the membrane structure. These changes are accelerated at higher olefin feed
pressures or higher silver salt concentrations. Previously (Figure 15), it was shown that Pebax®
2533+AgBF4 membranes sorb tremendous amounts of olefins. For example, at 50 psig ethylene
uptake in these membranes is approximately 25 volume percent. One hypothesis is that once
large amounts of olefin sorb, the highly swollen membrane allows redistribution of silver ions
that result in the formation of ionic clusters (BF4- and Ag+). Such clusters would be a barrier to
diffusion for all species, but might have a greater impact on olefins because redistribution of Ag+
could interrupt the proposed hopping mechanism of facilitated transport.
60
50
40
Mixed-gas
C2H4/C2H6
30
selectivity
7 % C2H4 feed
20
10
50 %
C2H4 feed 15 % C2H4 feed
0
0 50 100 150 200 250 300
Figure 26. Mixed-gas ethylene/ethane selectivity of Pebax® 2533 membranes containing 80

wt% AgBF4 as a function of permeation time for three different feed mixtures.
Tests were conducted at room temperature with 50 psig feed pressure and 0 psig
permeate pressure.
5.4 Investigating the Reversibility of Olefin Conditioning
To examine the reversibility of olefin-induced permeance decline, we conducted an experiment

where the flow of feed gas is removed from the membrane for a period of time and then
restarted. Figure 27 presents pure-gas ethylene permeance as a function of time for a standard
Pebax® 2533+AgBF4 membrane. For the first 2.3 hours, the stamp shows a typical decrease in
ethylene permeance. After this period of time, the feed gas flow was turned off and the permeate
36 311 Final report

DE-FC36-04GO14151
and residue valves closed. The pressure on the feed side was purged to atmospheric, so that the
membrane was universally exposed to 14.7 psia of dry ethylene. The stamp was then held under
this static condition for one hour. Thereafter, the feed flow was restarted, feed pressure set to 50
psig, valves opened, and permeation measurements resumed. After the one hour recovery time,
membrane ethylene permeance increased four times and then began to decline again. The
membrane did not recover completely to the initial permeance, but this may simply be related to
the kinetics of the recovery process. This recovery phenomenon may be responsible for previous
long-term results that appeared to show stable permeation behavior (see Figure 20, for example).
When the feed flow was periodically turned off to conserve gas, it may have allowed the
membrane permeance to recover in these tests. We believe that the limited long-term data
available in the literature for facilitated olefin transport membranes may be tainted by this effect.
Stamp
held at
0 psig
Pure-gas
10
ethylene
permeance
(gpu)
valves valves
closed opened
1
0 1 2 3 4 5
Figure 27. Pure-gas ethylene permeance as a function of permeation time for a standard
Pebax® 2533+80 wt% AgBF4 composite membrane. The feed pressure in this
experiment was 50 psig and the permeate pressure was 0 psig.
A number of experiments were conducted where membrane samples were periodically evacuated
or flows were reversed after olefin conditioning. Table 3 demonstrates the effect of a permeate-
side vacuum on facilitated transport membrane performance. For this experiment, after nitrogen
permeance was found to be very low (0.3 gpu), pure-gas ethylene permeance was monitored for
2 hours. During this time, ethylene permeance decreased from 46 to 2.6 gpu. At this point,
nitrogen permeance was re-measured, the feed pressure was turned off, and a vacuum was
applied to the permeate side of the membrane for 5 minutes. After vacuum treatment, nitrogen
permeance was measured again and ethylene permeance was monitored over time. The data
show that ethylene permeance increases and nitrogen permeance decreases after vacuum
37 311 Final report

DE-FC36-04GO14151
treatment. This result suggests that the olefin conditioning effect is at least partially reversible.
Nitrogen permeance probably decreases because the vacuum removes ethylene from the
membrane, allowing it to recover from a highly swollen state. Ethylene permeance recovers
after vacuum treatment presumably because silver ion-olefin clusters break up, allowing active
silver ions to return to something close to their original distribution. We repeated this cycle of
ethylene permeation and vacuum treatment five times, and in each case, ethylene permeance
increased while nitrogen permeance decreased or remained constant after vacuum exposure.
Finally, after 380 minutes (and six vacuum cycles), the nitrogen permeance increased
significantly, indicating that the membrane had developed defects.
Table 3. Effect of Vacuum Treatment on Pebax®+80 wt% AgBF4 Membrane Permeance
Permeation Ethylene Nitrogen

Time Permeance Permeance
(min) (gpu) (gpu)
0 46 0.3
30 23 -
60 12 -
90 7.3 -
120 2.6 1.2
Vacuum for 5 minutes
140 12 0.9
170 2.9 1.8
Vacuum cycle repeated 5 times
380 21 33
410 4.2 30
To further investigate the reversibility of olefin conditioning, we tried periodically reversing the
pressure gradient on the membrane and monitoring the effect on ethylene permeance. Table 4
summarizes the data obtained during this experiment.
38 311 Final report

DE-FC36-04GO14151
Table 4. Effect of Reverse-Flow Treatment on Pebax®+80 wt% AgBF4 Membrane

Permeance
Permeation Ethylene Nitrogen

Time Permeance Permeance
(min) (gpu) (gpu)
0 25 low
30 21
60 13
90 8.6
Flow 5 psig of ethylene on the permeate
face of the membrane for 5 minutes
110 9.7
125 11
Flow 5 psig of ethylene on the permeate
face of the membrane overnight
0 35 1.1
Flow cycle 3 times
255 5.1 20
Add polymer coating
330 48 0.7
Initially, the membrane permeated pure ethylene for 90 minutes until the permeance was less
than 10 gpu. The feed side of the cell was then degassed to 0 psig and 5 psig of ethylene was
applied to the permeate side of the membrane for 5 minutes. This short reverse-flow treatment
seemed to have little impact on membrane properties, so it was repeated for a longer interval of
time (overnight). After approximately 12 hours, when the flow was switched back to the normal
configuration, the ethylene permeance had increased to a value even higher than the initial
permeance. Apparently, reversing the olefin pressure gradient or simply reducing the olefin
concentration in the membrane for a significant amount of time allows the membrane structure to
revert back to a form that is more like that in the fresh, high-flux membrane.
Similarly to the vacuum exposure results, the membrane exhibited high nitrogen permeance after
several cycles of reverse-flow treatment, indicating defects had formed. The formation of
defects is probably caused by membrane expansion and contraction resulting from sorption and
desorption of ethylene. As mentioned previously, at 50 psig, ethylene uptake in Pebax® 2533
membranes containing 80 wt% AgBF4 is approximately 25 volume percent. This amount of
sorption must cause significant polymer dilation. It is certainly reasonable that the pressure
cycling described in Tables 3 and 4 could cause swelling and contraction that encourages
membrane defect formation. As a final test to determine whether the bulk of the membrane was
still functional, a thin coating of a highly-permeable polymer was applied to the membrane to
seal defects. Subsequent testing showed that the membrane nearly returned to initial
performance levels. This result demonstrates that in the absence of defects, the olefin
conditioning effect is reversible.
39 311 Final report

DE-FC36-04GO14151
5.5 FTIR Analysis of Olefin Conditioning
Previously, it has been shown that FTIR spectroscopy can be used to study the interactions of
silver ions dispersed in a polymer matrix. We have employed this technique in an effort to
understand the membrane changes that occur after olefin permeation. Figure 28 presents FTIR
scans for PEO membranes containing 0, 40, and 80 wt% AgBF4. Both PEO and Pebax® 2533
contain ether linkages in the polymer backbone that are capable of interacting with silver ions. It
is believed that these interactions allow silver ions to remain disassociated in the polymer matrix,
and thus active for olefin complexation. The peak corresponding to ether linkages is evident for
pure PEO at approximately 1100 cm-1. When AgBF4 is added to PEO, the ether peak broadens
and shifts to lower wave numbers (shifts to the right in the figure). This result is consistent with
previous data for silver salt facilitated transport membranes, and indicates that polymer C-O
bonds are weakened due to the interaction of ether oxygens with Ag+ ions.
PEO
PEO + 40 wt% AgBF4 1102
%T
PEO + 80 wt% AgBF4

1033
1037
1550.0 1500 1450 1400 1350 1300 1250 1200 1150 1100 1050 1000 950 900.0
cm-1
Figure 28. FTIR spectrographic scans of PEO membranes containing 0, 40, and 80 wt%
AgBF4. %T = Relative percent transmission.
Figure 29 shows FTIR scans for three membranes containing 80 wt% AgBF4. One membrane is
a fresh, as-cast sample, while the other two are aged membranes – one aged by exposure to the
reducing gas hydrogen and the other aged by long-term ethylene permeation. The scans for the
two aged membranes are very different. The hydrogen-exposed sample is black, indicating the
presence of a large number of reduced Ag0 particles. In contrast, the ethylene-exposed sample
appears nearly white, similar to the fresh, as-cast sample. This visual difference is mirrored in
the FTIR results of Figure 29. The hydrogen-aged sample does not show much of a peak in the
1000-1100 cm-1 range, indicating little interaction of Ag+ ions with ether oxygens, and consistent
with the notion that Ag+ ions have been reduced to Ag0 particles. In fact, the distinct absence of
peaks in the entire range of the hydrogen-aged sample scan suggests that the large number of
silver metal particles attenuate the FTIR signal altogether. In contrast, the ethylene permeation
sample still shows a discernable ether linkage peak, indicating that at least some active Ag+ ions
40 311 Final report

DE-FC36-04GO14151
are present. This result adds to the evidence that two separate performance degradation
mechanisms are operative in our facilitated transport membranes: in addition to the well-known
carrier poisoning or reduction issue, the membranes show apparent olefin-induced performance
loss that occurs even though active Ag+ ions are present in the membrane.
PEO + 80 wt% AgBF4 (Fresh)
PEO + 80 wt% AgBF4 (Aged by hydrogen exposure)
%T
PEO + 80 wt% AgBF4 (Aged by ethylene permeation)
1788.8 1700 1600 1500 1400 1300 1200 1100 1000 900 775.7
cm-1
Figure 29. FTIR spectrographic scans of PEO membranes containing 80 wt% AgBF4 that
were freshly prepared, aged by hydrogen exposure, or aged by ethylene
permeation. %T = Relative percent transmission.
5.6 Effect of Temperature, Polymer Plasticizers and Humidified Feed
Effect of Temperature
To further investigate the nature of olefin-induced performance decline in our membranes, we

studied the effect of temperature on this phenomenon. Figure 30 presents normalized ethylene
and ethane mixed-gas permeances in Pebax® 2533 containing 80 wt% AgBF4 as a function of
time and temperature. Based on the data in Figure 30(a), mixed-gas ethylene permeance
decreases faster as temperature is increased. For example, after one hour of operation, ethylene
permeance is reduced to 65% of its initial value at 5°C, while at 40°C the corresponding
permeance is only 5% of the starting value. The impact of temperature on ethane permeance
decline [Figure 30(b)] is less clear, predominately because the very low permeances (<2 gpu) for
this non-facilitated gas make accurate measurements difficult. Nevertheless, certainly for
ethylene and perhaps for ethane as well, permeance decline occurs faster at higher temperatures
where the membrane matrix is more mobile. Later in the discussion in this section, we show that
plasticizers, which increase matrix mobility, also accelerate permeance decline. These results
are consistent with the notion that olefin-induced structural changes in the membrane, such as
movement or clustering of silver ions, are responsible for the time-dependent membrane
behavior.
41 311 Final report

DE-FC36-04GO14151
1 1
Normalized ethylene permeance (a) (b)
Normalized ethane permeance

0.8 0.8
0.6 5 °C 0.6
0.4 0.4
22 °C 22 °C
0.2 0.2 5 °C
40 °C 40 °C
0 0
0 50 100 150 0 50 100 150
Permeation time (min) Permeation time (min)
Figure 30. Normalized mixed-gas (a) ethylene and (b) ethane permeances of Pebax® 2533
membranes containing 80 wt% AgBF4 as a function of time and temperature.
The data in Figure 30 indicate that lowering the system temperature is a way to slow the rate of
ethylene permeance decline in facilitated transport membranes. Figure 31 shows that lower
temperature also improves the stability of membrane mixed-gas selectivity. At 40°C, the
ethylene/ethane selectivity is initially nearly 50, but decreases quite rapidly with increasing
permeation time. At 5°C, the membrane behaves quite differently: selectivity starts off at only
20, but then actually increases over the course of the experiment. The selectivity increase occurs
because ethane permeance decreases faster over time than that of ethylene. The reason for this
behavior is not readily apparent; however, it clearly implies that there may be benefits to running
the facilitated transport membranes at low temperature. This intriguing result warrants more
long-term study.
42 311 Final report

DE-FC36-04GO14151
50
45
40
35
Mixed-gas
C2H4 / C2H6
30
selectivity 5 °C
22 °C
25
20
40 °C
15
10
0 50 100 150
Permeation time (min)
Figure 31. Mixed-gas ethylene/ethane selectivity of Pebax® 2533 membranes containing 80

wt% AgBF4 as a function of time and temperature.
Effect of Polymer Plasticizers
One theory proposed to describe the olefin-induced permeance decline described in this section
is that the silver carrier ions are mobile within the polymer matrix. When the membrane is
exposed to an olefin, the polymer matrix becomes highly swollen as olefin sorbs into the
membrane to react with silver ions. In this mobile environment, silver ions may cluster, as they
are known to do in liquid solutions. These crystal-like clusters would represent low permeability
regions, through which olefins and non-facilitated gases would have a difficult time passing.
This type of structural rearrangement might also explain why facilitated transport membranes
often become defective during long-term olefin permeation tests. Incompatibility between rigid
crystal-like domains and polymeric regions could yield micron-scale interfacial regions where
gases flow non-selectively.
To see whether these low-permeability regions could be broken up, we added a plasticizer to our
baseline facilitated transport membrane. Polyethylene glycol (PEG) with a molecular weight of
200 is a non-volatile liquid that is compatible with Pebax® 2533 and silver salts. A number of
early attempts failed to yield defect-free membranes; however, we eventually succeeded in
fabricating membranes containing 80 wt% AgBF4 in a matrix of 50% PEG and 50% Pebax®
2533. Figure 32 compares mixed-gas ethylene permeance as a function of time in this
membrane, a baseline Pebax® 2533+80 wt% AgBF4 membrane, and a control membrane without
silver salt. Interestingly, ethylene permeance decreases faster over time in the Pebax®
2533+AgBF4 membrane containing PEG, relative to the baseline membrane without plasticizer.
43 311 Final report

DE-FC36-04GO14151
It appears that rather than enhancing membrane permeances, the PEG plasticizer accelerates
structural changes in the membranes that cause permeance decline. This result is consistent with
the temperature effect data, which shows that increased matrix mobility increases the rate of
permeance decline.
®
Pebax 2533
10
Mixed-gas
ethylene
®
permeance Pebax 2533 +
(gpu) 80 wt% AgBF4
1
®
Pebax 2533 + PEG (1:1) + 80 wt% AgBF4
0 24 48 72 96
Figure 32. Mixed-gas ethylene permeance as a function of permeation time for plasticized
and unplasticized Pebax® 2533 membranes.
Effect of Humidified Feed
In prior testing, we have shown that membranes containing less than 40 wt% AgBF4 exhibit
stable permeation properties, although selectivities are low. Water vapor is known to increase
olefin permeation and selectivity in facilitated transport membranes that use silver salts.
Consequently, experiments were conducted with a humidified feed gas to see whether water
vapor would enhance the performance of membranes containing low amounts of silver salt.
Figure 33 shows mixed-gas ethylene permeances as a function of permeation time for PEO
membranes containing 20, 40, and 60 wt% AgBF4 with a humidified feed. Gas chromatograph
measurements showed that the feed mixture contained about 0.15 mol% water vapor, which
amounts to an activity of approximately 0.3 at room temperature.
44 311 Final report

DE-FC36-04GO14151
100 (a) 100 (b)

60 wt% AgBF4 60 wt% AgBF4
Mixed-gas C2 selectivity
Ethylene permeance (gpu)
40 wt% AgBF4
40 wt% AgBF4
10
20 wt% AgBF4
10
20 wt% AgBF4
1
0 10 20 30 40 50 0 10 20 30 40 50
Figure 33. Mixed-gas (a) ethylene permeance and (b) ethylene/ethane selectivity for PEO
composite membranes containing 20, 40, and 60 wt% AgBF4 with a humidified
feed gas. Water content of the feed gas was 0.15 mol% as determined by gas
chromatography.
Initially, the ethylene permeances and selectivities of the membranes with a humidified feed
were very high. For example, the initial mixed-gas ethylene/ethane selectivity for a membrane
containing 20 wt% AgBF4 was approximately 10, compared to a typical value of 1.5 for this
same membrane with a dry feed. Despite the high initial performance, the membrane
permeances and selectivities decreased continuously over time. After 10 days, the ethylene
permeance and ethylene/ethane selectivity of the 20 wt% AgBF4 membrane were less than half
the initial values. Similar results were found for the other silver salt concentrations, with the 60
wt% AgBF4 membrane showing the most rapid decline in performance. This result is consistent
with the idea that higher amounts of silver salt cause more pronounced time-dependent
permeation behavior, presumably due to membrane structural changes induced by the
complexation reaction of the carrier with olefin molecules.
The time-dependent permeation behavior shown in Figure 33 for a humidified feed with
membranes containing as little as 20 wt% AgBF4 is different from results using a dry feed. For
example, with a dry feed, a membrane containing 20 wt% AgBF4 shows nearly constant ethylene
permeances and ethylene/ethane selectivities of <10 gpu and 1.5, respectively (see Figure 18).
These values are much less than the initial performance with a humidified feed. However, after
48 days, the performance of the humidified 20 wt% AgBF4 membrane appears to be approaching
that of the typical dry membrane permeance and selectivity. While water vapor definitely
enhances the initial performance of membranes with low silver salt content, it does not solve the
membrane stability issue. It appears that a membrane with useful olefin/paraffin selectivities is
subject to olefin-induced performance decline, regardless of whether it is a dry membrane with
>40 wt% AgBF4 or a humidified membrane with lower silver salt content.
45 311 Final report

DE-FC36-04GO14151
After completing the long-term study shown in Figure 33, sample membranes prepared at the
beginning of the test and stored at ambient conditions during the experiment were examined.
Figure 34 shows a picture of these stored membranes after 48 days, as well as the membranes run
in the long-term, humidified feed test.
a) Membranes run in system b) Stored membranes
20
20 wt%
AgBF44
40 wt%
40
AgBF
AgB 4 4
60
60 wt%
AgB
AgBF44
Figure 34. Photograph of (a) membranes run in the humidified feed permeation system and
(b) stored membranes after 48 days.
Originally, all of the membranes were white; however, after 48 days, the stored membranes (b)
have turned brown or black, indicating silver particle reduction. The color of the membranes
follows their initial AgBF4 concentration, with higher silver content yielding darker membranes.
A similar trend with silver content is observed for the membranes run in the test system.
Compared to the stored membranes, the test system samples (a) are much lighter in color,
suggesting less silver reduction. This is consistent with the fact that the test system samples
were shielded from light in the permeation cell, while the stored samples were exposed to
ambient lighting (which is known to reduce silver ions).
After 48 days, the stored membranes were inserted into the re-circulating permeation system and
tested with the humidified ethylene/ethane feed. The test results for these stored membranes, as
well as the original and aged membranes, are summarized in Table 5. Interestingly, even though
the stored membranes exhibit signs of greater silver reduction (see Figure 34), they show higher
ethylene permeances and ethylene/ethane selectivities than the membranes that ran in the system
for 48 days. In fact, the 20 and 40 wt% AgBF4 membranes that were stored for 48 days show
similar performance to fresh, as-cast membranes. This result again indicates that silver carrier
reduction is not the only mechanism of performance decline. In fact, in this case, 48 days of
olefin exposure has a significantly more negative impact on membrane performance than the
same period of carrier reduction by light exposure.
46 311 Final report

DE-FC36-04GO14151
Table 5. Properties of Aged PEO Membranes Containing 20 to 60 wt% AgBF4:

Comparison of Stored and Tested Samples after 48 Days.
AgBF4 Content Mixed-gas Permeance (gpu) Mixed-gas C2

Membrane Type
wt % C2H4 C2H6 Selectivity
20 19 2.1 9.0
Fresh Membranes:
40 22 1.5 15
Initial Properties
60 79 0.8 99
20 6.9 4.1 1.7
Aged Membranes:
After 48 Days Of 40 10 2.3 4.3
Testing
60 9.1 4.5 2.0
20 26 2.5 10
Stored
Membranes: 40 38 2.5 15
48 Days Old
60 43 2.3 19
5.7 Sorption Hysteresis
In an effort to better understand the fundamentals of the silver ion-olefin complexation reaction,
we conducted sorption studies on our polymer + silver-salt selective layer material. Figure 35
presents some initial results in the form of an ethylene sorption/desorption isotherm for Pebax®
2533 containing 50 wt% AgBF4. For this experiment, a film of Pebax® 2533 + AgBF4 was cast
from an ethanol solution and dried at ambient conditions. The sample was then inserted into a
barometric sorption apparatus and exposed to vacuum for more than 24 hours. This drying
protocol removes residual solvent, water and air gases from the sample. The dried sample was
then exposed to progressively higher ethylene pressures and the equilibrium sorption uptake was
measured at each pressure stage.
47 311 Final report

DE-FC36-04GO14151
250
desorption
Ethylene concentration, cm3 (STP)/cm3

200
first sorption
150
second sorption
100
50
Pebax® 2533 + 50 wt% AgBF4

0
0 2 4 6 8 10 12 14 16
Ethylene pressure (atm)
Figure 35. Ethylene sorption and desorption isotherms at 25°C in Pebax® 2533 containing 50
wt% AgBF4.
• The curve labeled “first sorption” in Figure 3 represents the initial ethylene uptake upon
first exposure of the polymer sample to the gas. The amount of ethylene sorption in the
Pebax® 2533+50 wt% AgBF4 sample is enormous. For example, at 10 atm this sample
sorbs nearly 200 cm3(STP)/cm3 of ethylene, compared to only about 10 cm3(STP)/cm3 in
pure Pebax® 2533. This 20-fold increase in ethylene sorption after adding AgBF4 is
clearly due to the presence of silver ions that complex with ethylene.
• Interestingly, it appears that this complexation reaction is not completely reversible. The
open points labeled “desorption” in Figure 35 represent the amount of ethylene that
remains in the sample as the ethylene pressure is lowered. Typically, gas
sorption/desorption in polymer samples is completely reversible and the desorption
isotherm falls on top of that measured for sorption. This is not the case for Pebax® 2533
containing 50 wt% AgBF4, where hysteresis is clearly evident. As the gas pressure is
lowered, the sample releases ethylene very slowly. For example, at 3 atm, the amount of
ethylene sorbed in the sample during first sorption and desorption is roughly 110 and 170
cm3(STP)/cm3, respectively. This result suggests that after complexation occurs, silver
ions in our polymer system hold onto ethylene molecules and do not disassociate as
readily as was believed.
48 311 Final report

DE-FC36-04GO14151
• After the final desorption point in Figure 35, there still appeared to be more than 130
cm3(STP)/cm3 of ethylene sorbed in the Pebax 2533/AgBF4 sample. The sample was
then exposed to a vacuum overnight before beginning a second sorption isotherm. There
is no way to tell from this experiment whether all of the sorbed ethylene was removed
during the vacuum treatment. It is possible that some of the ethylene remained in the
sample bound to silver ions.
• The second ethylene sorption isotherm (labeled “second sorption” in Figure 35) shows
significantly lower ethylene uptake relative to the first isotherm. This suggests that some
of the silver ions are not available to complex ethylene during the second sorption run.
Apparently, some silver ions have become inactivated either because they are already
bound to ethylene from the first run or through some other structural change. Future
work will try to determine whether the polymer-silver system continues to change with
repeated olefin cycling.
These sorption findings (i.e., hysteresis) are consistent with the time-dependent membrane
properties described throughout this section. Clearly, olefin permeation in a silver salt facilitated
transport membrane is not a simple idealized case of olefin molecules moving through an inert
matrix, reversibly complexing with fixed carrier ions. Rather, it appears that olefin complexation
by silver ions may be only partially reversible over short timescales (i.e., the desorption step is
relatively slow) and sorption swells the matrix, creating a mobile environment that allows
significant membrane structural changes over time.
5.8 Summary of Time-Dependent Membrane Studies
We have isolated a new mechanism of performance decline in silver salt facilitated transport
membranes that is different from conventional carrier reduction or poisoning phenomena
associated with facilitated transport membranes employing silver salts. Conventional carrier
poisoning results in a loss of active silver ions in the membrane. This mechanism causes
membranes to turn black, yields an attenuated and unshifted FTIR peak, and increases non-
facilitated gas permeances. In contrast, the olefin-induced performance decline we have
identified in this work lowers non-facilitated gas permeance in mixtures with olefins and does
not result in FTIR signal or membrane color changes. It appears that regardless of the polymer
matrix or type of silver salt used, gas permeance begins to decrease once the membrane has been
exposed to an olefin. This effect is accelerated at higher temperatures and in the presence of
water or polymeric plasticizers, suggesting a structural change to the membrane resulting from
the silver ion-olefin complexation reaction. The mechanism of olefin-induced permeance
decline warrants further investigation.
6. CARRIER POISONING AND REGENERATION OF SILVER SALT

FACILIATED TRANSPORT MEMBRANES
6.1 Carrier Poisoning Studies
Previous work by others on facilitated transport membranes rarely mentions the impact of carrier
poisoning on membrane performance. In fact, to our knowledge, there is no published
49 311 Final report

DE-FC36-04GO14151
information quantitatively describing the effect of poisoning species on the separation

performance of facilitated olefin transport membranes. As mentioned earlier, the principal
species believed to interact with silver ions, and which might be expected to be present in
industrial gas streams, are hydrogen, hydrogen sulfide, and acetylene. Consequently, we have
investigated the effect of exposure to these compounds over time on membrane performance.
We have also examined the effect of light on membrane separation performance because silver
salts are known to be light sensitive.
Figure 36 presents the mixed-gas fluxes and ethylene/ethane selectivity of Pebax® 2533+80 wt%
AgBF4 facilitated transport membranes as a function of aging time in ambient light. It is well
known that silver ions are gradually reduced to silver metal upon exposure to light.[39] In fact,
this is the principle behind chemical-based photography. The greater the exposure to light, the
more silver ions are reduced to silver metal nanoparticles, which appear dark in color. Over the
course of 34 days, the exposed facilitated transport membrane progressively turned darker until it
was completely black at the end of the experiment. Correspondingly, the mixed-gas selectivity
of the membrane decreased from 34 to 1.1 as the silver carrier ions became deactivated. From
the flux data, it is evident that the ethane/ethylene selectivity loss was largely due to increased
ethane flux over time. If one examines the effect of silver ion content on gas fluxes from Figure
2 (Section 2), the observed aging effects can be explained as traveling from right to left on this
plot. In other words, as light reduces silver from ion form to metal, we are moving from high
Ag+ content (the right side of Figure 2) to low Ag+ content (the left side of Figure 2). In doing
so, ethane flux should increase monotonically (as it does in Figure 36), while ethylene flux
should decrease initially and than increase slightly (as it does in Figure 36).
100 100
50/50 C2 mixture
50 psig feed
room temperature
Mixed-gas flux (gpu)
Ethylene
10
10
Ethane
1
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Aging time (days) Aging time (days)
Figure 36. Mixed-gas ethylene and ethane fluxes and ethylene/ethane selectivity as a function
of aging time in ambient light for a Pebax® 2533+80 wt% AgBF4 facilitated
transport membrane.
50 311 Final report

DE-FC36-04GO14151
The detrimental impact of light on our facilitated transport membranes is not considered a
serious operational problem due to the fact that the membrane is not exposed to light once it is
fabricated into modules. However, the effect of various chemical species on membrane
performance is a concern. Figure 37 presents mixed-gas ethylene and ethane permeance and
selectivity for Pebax® 2533+AgBF4 membranes that were exposed to hydrogen between
measurements. In these tests, the membrane continuously permeated pure hydrogen at 20 psig
feed and 0 psig permeate pressure except for periodic interruptions where mixed-gas ethylene
and ethane fluxes were measured. Hydrogen is a reducing gas and is believed to react with silver
ions in the following way:
2Ag + + H 2 ⎯⎯⎯
→ 2Ag 0 + 2H +
The data in Figure 37(b) show that ethylene/ethane selectivity decreases rather rapidly after three
days, and after a week is only slightly greater than one. Figure 37(a) indicates this loss of
selectivity is primarily due to increasing ethane (non-facilitated gas) permeance. Such behavior
is consistent with a decrease in the ionic character of the membrane over time. This transport
result is supported by a visual inspection of the membranes. The membranes, which are initially
white, turn black after hydrogen exposure, indicating that silver ions have been reduced to silver
particles (Figure 38).
1,000 100
Pebax ® 2533+AgBF4 (1:4) (a) (b)
100
Ethylene
10
10
Ethane
1 1
0 2 4 6 8 0 2 4 6 8
H2 permeation time (days) H2 permeation time (days)
Figure 37. Mixed-gas ethylene and ethane (a) permeance and (b) selectivity as a function of
hydrogen exposure time for Pebax® 2533+AgBF4 composite membranes.
51 311 Final report

DE-FC36-04GO14151
2Ag + + H 2 ←
⎯→ 2 Ag 0 + 2 H +
Figure 38. Picture of a Pebax® 2533+AgBF4 composite membrane stamp before (white) and
after (black) hydrogen permeation.
This hydrogen exposure experiment is a particularly harsh test because the membrane permeates
pure hydrogen continuously between mixed-gas ethylene/ethane measurements. In the actual
applications where the membrane would be used, such as at the bottom of a C2 splitter or on
metathesis C4 purge streams, the concentration of hydrogen is in the ppm range according to our
ABB Lummus colleagues. Additionally, in a mixture with olefins, silver ions may be more
resistant to hydrogen reduction because they are complexed with olefin molecules. Nevertheless,
it is clear that hydrogen does reduce impregnated silver ions in our facilitated transport
membranes, thereby degrading performance over time.
Figure 39 presents the effect of hydrogen sulfide exposure on mixed-gas Pebax® 2533+AgBF4
membrane performance. During this experiment, the membrane continuously permeated 10 ppm
of H2S in ethylene between ethylene/ethane measurements. Similar to the results with hydrogen
poisoning (Figure 37), membrane mixed-gas ethylene/ethane selectivity decreases dramatically
over one week. In contrast to the hydrogen results, for H2S exposure, the permeance of both
ethylene and ethane decrease over time. This result may reflect the formation of impermeable
silver sulfide crystals in the membrane. It is believed that H2S reacts with silver ions by the
following mechanism:
H 2S + 2AgBF4 ⎯⎯⎯
→ Ag 2S + 2HBF4
At the end of the hydrogen sulfide exposure experiment, the membrane was reddish-brown in
color (see Figure 40). Previously, we calculated that for a 10 ppm H2S feed, if every hydrogen
sulfide molecule participated in the silver sulfide formation reaction, all of the silver ions in the
membrane would be converted to Ag2S in roughly five days. Based on the data in Figure 39, this
estimate appears reasonably accurate, given that after seven days there is very little facilitation of
ethylene. Like the hydrogen exposure experiment, this hydrogen sulfide study is a particularly
harsh test of membrane durability.
52 311 Final report

DE-FC36-04GO14151
1,000 100
(a) (b)
Mixed-gas C2H4C2H6 selectivity

100
Ethylene
10 10
Ethane
0.1 1
0 2 4 6 8 0 2 4 6 8
H2S permeation time (days) H2S permeation time (days)
Figure 39. Mixed-gas ethylene and ethane (a) permeance and (b) selectivity as a function of
hydrogen sulfide permeation time for Pebax® 2533+80 wt% AgBF4 membranes.
The membrane samples continuously permeated 10 ppm H2S in ethylene between
mixture permeation measurements.
H 2S + 2AgBF4 ⎯
⎯→ Ag 2S + 2HBF4
Figure 40. Picture of a Pebax® 2533+80 wt% AgBF4 composite membrane stamp before
(white) and after (brown) hydrogen sulfide permeation.
Effective sulfur clean-up technologies routinely reduce H2S levels in gas streams to <1 ppm.
Based on discussions with ABB Lummus, sulfur species in the applications where our
membranes would be used (bottoms of C2 splitter, metathesis C4 purge) are typically less than
0.1 ppm. This is two orders of magnitude lower than the amount tested in Figure 39. We did not
have time to quantify the effect of lower H2S concentrations on membrane performance in this
program. It is reasonable to expect that lower contaminant concentrations would result in longer
membrane lifetimes.
Pebax® 2533+80 wt% AgBF4 facilitated transport membrane was also tested with a feed gas
containing 10 ppm of acetylene (C2H2) in ethylene at 20 psig. Similarly to results presented for
53 311 Final report

DE-FC36-04GO14151
the other poisoning species, membrane selectivity decreased significantly after acetylene
exposure. For instance, after five days of permeating acetylene, the mixed-gas ethylene/ethane
selectivity of a membrane stamp was reduced from 32 to 4.7. At the end of the five days, the
stamp was removed from the test cell and examined. In contrast to a control which appeared to
be in the as-cast condition, the acetylene-exposed sample had turned dark brown. As in the case
of results obtained for light and hydrogen exposure, the loss in selectivity for the acetylene-
exposed membrane resulted primarily from an increase in ethane flux. This is in contrast to the
hydrogen sulfide exposure tests where both ethylene and ethane fluxes decreased tremendously.
As will be shown in the following section, this difference in exposure impact on transport
properties also manifests itself as a difference in the reversibility of the poisoning reactions.
While it seems that carrier deactivation induced by light, hydrogen and acetylene can be reversed
under simple regeneration conditions, the effect of hydrogen sulfide on the silver carrier appears
to be irreversible.
6.2 Stabilization and In-situ Regeneration of Reduced Silver Salt Membranes
A key technical challenge identified at the start of this project was to improve the lifetime of our
facilitated transport membranes in the presence of process stream components or conditions that
may deactivate the olefin carrier species. The work in the previous section has shown that
active, olefin-complexing Ag+ ions can be deactivated by reduction to silver metal (Ag0), caused
by exposure to light or reducing chemical species such as hydrogen, and by reaction with
hydrogen sulfide (H2S) or acetylene (C2H2) to yield non-complexing silver compounds (Ag2S or
Ag2C2). Once Ag+ ions have been reduced, they are no longer active olefin carriers and
membrane performance deteriorates.
During the course of these studies, we discovered that almost all aged and some poisoned Pebax®
2533+AgBF4 membranes can be regenerated by exposure to an oxidizing vapor or liquid
treatment. This discovery offers the possibility that membrane lifetime could be enhanced by a
periodic, in-situ regeneration treatment, much like ultra- or microfiltration membranes are
regularly backflushed to manage fouling.
The finding that facilitated olefin transport membranes can be regenerated evolved from an effort
to stabilize our Pebax® 2533+AgBF4 coating solutions. Without the addition of a stabilizing
agent, coating solutions turn black within a few hours, indicating that silver ions are being
reduced to silver metal particles. After a week, such reduction leads to large silver deposits on
the container walls. This ‘plate-out’ occurs more rapidly if the solution is exposed to light, but
proceeds even in a dark environment, suggesting that the polymer, solvent, and/or container
walls contribute to silver reduction. Based on patent literature from the 1970s,[32] we added
small amounts of hydrogen peroxide (H2O2) and hydrogen borofluoride (HBF4) to our coating
solution to prevent silver ion reduction. This strategy proved very effective, as H2O2/HBF4
additions of less than 0.5 wt% completely stabilized the coating solutions. It is believed that the
strong oxidizing agent, H2O2, can react in an acidic environment with silver metal to yield silver
ions in the following manner:
+
H 2 O 2 + Ag 0 ⎯⎯
H
→ Ag + + 2H 2 O + H 2
54 311 Final report

DE-FC36-04GO14151
Hydrogen peroxide is known to be quite unstable at pH levels greater than 5, and testing
confirmed that without the addition of an acid, the stabilizing effect of H2O2 on the coating
solution is rather short-lived. An initial screening suggested that HBF4 in approximately a 1:1
ratio with H2O2 was most effective at stabilizing the coating solution and restoring aged
membrane properties.
Figure 41 illustrates the impact of H2O2/HBF4 stabilizer added to Pebax® 2533+AgBF4 coating
solutions. The coating solutions in the containers were more than nine months old at the time of
this picture. The solution without stabilizer is black and membranes prepared from it are brown.
In contrast, the stabilized solution is tan and membranes prepared from it appear white (actually
the selective layer is transparent and the white appearance comes from the microporous support
membrane). Table 6 presents the transport properties of membranes prepared with and without
stabilizer in the solution. The permeances and selectivities of the membrane stamps are similar.
This result suggests that despite the dark color of the unstabilized solution and membranes, there
are still enough active silver ions in these membranes to enable olefin facilitation. Thus,
although the H2O2/HBF4 stabilizer prevents coating solution plate-out, membranes with good
initial properties can be coated from aged solutions with or without the stabilizer.
Figure 41. Comparison of Pebax® 2533+80 wt% AgBF4 coating solutions prepared with and
without H2O2/HBF4 stabilizer. Both solutions are >9 months old. The container
on the right contains 0.5 wt% H2O2/HBF4. In the foreground are composite
membranes as they appear immediately after coating PVDF support with the
respective solution.
55 311 Final report

DE-FC36-04GO14151
Table 6. Comparison of Pebax® 2533+80 wt% AgBF4 Membranes Prepared from Aged
Solutions With and Without H2O2/HBF4 Stabilizer
Mixed-Gas Permeance Mixed-Gas

Membrane (gpu)
Membrane Description C2H4/C2H6
Appearance
C2H4 C2H6 Selectivity
Membrane prepared from old 79 2.4 33

solution without H2O2/HBF4
66 1.7 39
49 1.4 35
Membrane prepared from old
solution with H2O2/HBF4
54 1.3 41
The coating solution with stabilizer contained 0.5 wt% H2O2 aqueous solution (35 wt% in H2O) and 0.5 wt% HBF4 aqueous solution
(50 wt% in H2O). Both coating solutions were >9 months old at the time membranes were prepared and tested. Test conditions: feed
pressure = 50 psig, T = 23°C. Mixture composition: 65% C2H4, 35% C2H6.
The situation is different for aged membranes coated from these two solutions. Table 7 presents
the transport properties and appearance of Pebax® 2533+AgBF4 membranes coated from
solutions with and without stabilizer and then aged for at least 9 months. After coating, the
membranes were stored in a sealed plastic bag. The membranes without H2O2/HBF4 stabilizer
have undergone significant silver reduction, as evidenced by their black color. Their transport
properties reflect this fact; the mixed-gas ethylene/ethane selectivities are less than four. In
contrast, aged membranes prepared from coating solution containing the H2O2/HBF4 stabilizer
are white and show mixed-gas ethylene/ethane selectivities of about 40, very close to the initial
values. Apparently, the presence of an oxidizing stabilizer in the coating solution allows for
stable storage of Pebax® 2533+AgBF4 membranes in a closed environment (plastic bag). The
beneficial effect is minimized if the membranes are stored in the open atmosphere (membrane is
noticeably stabilized for only 1-3 days). Presumably, the volatile stabilizer components
evaporate over time (these compounds are relatively low boiling liquids - Tb = 150°C and 130°C
for H2O2 and HBF4, respectively).
Table 7. Comparison of Aged Pebax® 2533+80 wt% AgBF4 Membranes Prepared from
Solutions With and Without H2O2/HBF4 Stabilizer
Mixed-Gas Permeance Mixed-Gas

Membrane (gpu)
Appearance
Nine month old membrane: 16 6.2 2.5

Without H2O2/HBF4
13 3.7 3.7
29 0.8 39
Nine month old membrane:
With H2O2/HBF4
22 0.5 44
The coating solution with stabilizer contained 0.5 wt% H2O2 aqueous solution (35 wt%) and 0.5 wt% HBF4 aqueous solution (50
wt%). Membranes were stored in a sealed plastic bag for roughly 9 months prior to testing. Test conditions: feed pressure = 50 psig,
T = 23°C. Mixture composition: 65% C2H4, 35% C2H6.
56 311 Final report

DE-FC36-04GO14151
In addition to stabilizing the coating solution and stored membranes, H2O2/HBF4 mixtures can
restore the properties of aged or poisoned Pebax® 2533+AgBF4 membranes. Table 8 presents
the transport properties and appearance of aged membranes (the same 9-month-old samples
without stabilizer from Table 7) before and after exposure to H2O2/HBF4 vapors. The membrane
color has changed from black to white and mixed-gas ethylene/ethane selectivities have
increased from less than 4 to about 20. This result is the first documentation of in-situ
regeneration of a facilitated olefin transport membrane.
Table 8. Regeneration of Aged Membrane Using H2O2/HBF4 Vapors
Mixed-Gas Permeances Mixed-Gas

Membrane (gpu)
Appearance
16 6.2 2.5
Old membrane
aged nine months
13 3.7 3.7
59 3.0 20
Old membrane
after regeneration with
H2O2/HBF4 vapors
40 2.5 16
Membrane was stored in a plastic bag for roughly 9 months prior to testing. Test conditions: feed pressure = 50 psig, T = 23°C.
Mixture composition: 65% C2H4, 35% C2H6. Regeneration conditions: Aged membrane sample was placed in a sealed container
that held a 50/50 mixture by weight of H2O2 aqueous solution (35 wt%) and HBF4 aqueous solution (50 wt%). The membrane was
in contact with the vapor pressure of this mixture at room temperature for ~16 hours.
From the aforementioned results, it is evident that peroxide/acid treatment can restore the
transport properties of photo-reduced membranes. However, the utility of this discovery is
significantly enhanced if the treatment can be shown to rejuvenate membranes deactivated by the
chemical poisoning species discussed in Section 6.1. Subsequent testing showed that facilitated
transport membranes deactivated by hydrogen and acetylene can be regenerated by peroxide/acid
treatment, while those poisoned by hydrogen sulfide cannot be restored. As an example, Figure
42 presents results for membrane stamps deactivated by hydrogen poisoning. As for light-
exposed membrane samples, treatment of a hydrogen-reduced stamp with H2O2/HBF4 vapors
results in a dramatic color change from black to white, and more importantly, a restoration of
high mixed-gas ethylene/ethane selectivity. Interestingly, the time required to restore the
hydrogen-reduced sample using H2O2/HBF4 vapors (~60 h) is considerably longer than that
required to regenerate light-reduced membranes (~10 h). The reason for this difference is
unknown, although it may be related to differences in the extent to which silver ions are reduced
throughout the bulk of the membrane.
57 311 Final report

DE-FC36-04GO14151
100
~ 60 h exposure to
acid/peroxide vapors
Mixed-gas
C2H4/C2H6
selectivity 10
1
0 3 6 9
H2 permeation time (days)
3 days 7 days 16 h exposed to 60 h exposed to

permeating H2 permeating H2 H2O2/HBF4 vapors H2O2/HBF4 vapors
Mixed-gas C2H4 / C2H6 results:

α = 28 α = 1.5 α = 3.2 α = 21
Figure 42. Deactivation of a Pebax® 2533+80 wt% AgBF4 composite membrane stamp by
hydrogen reduction and regeneration of the same stamp by H2O2/HBF4 vapor
treatment.
Peroxide/acid treatment was also effective at regenerating membrane samples deactivated by

acetylene exposure. For example, after five days of permeating 10 ppm acetylene, a membrane
stamp exhibited a mixed-gas ethylene flux of 29 gpu and an ethane flux of 6.3 gpu, yielding a
selectivity of 4.7. This same stamp, after 1 min immersion in a 50/50 H2O2/HBF4 bath,
demonstrated a mixed-gas selectivity of 22 based on an ethylene flux of 35 gpu and an ethane
flux of 1.6 gpu. In contrast, neither liquid nor vapor H2O2/HBF4 treatment produced any change
in the appearance or transport behavior of samples deactivated by hydrogen sulfide, even after
prolonged immersion periods (>1 week). At least with regard to the examined chemistry, the
reaction of hydrogen sulfide with Ag+ appears to be irreversible.
58 311 Final report

DE-FC36-04GO14151
Despite the inability to restore sulfur-poisoned facilitated olefin transport membranes with
H2O2/HBF4 vapors, the regeneration method shows promise as a means of prolonging membrane
lifetime. One might imagine a process whereby partially deactivated facilitated transport
modules could be periodically taken offline and regenerated by a peroxide/acid vapor stream. In
this manner, the regeneration process, in combination with sulfur removal upstream of the
membranes, could be a route to industrial use of facilitated olefin transport membranes. Because
of the need to protect species-sensitive catalysts, there are refinery olefin/paraffin streams that
are relatively clean, containing only small amounts of sulfur species. For example, based on
discussions with ABB Lummus, several applications where facilitated transport membranes
could be used (bottoms of C2 splitter, metathesis C4 purge) typically contain less than 0.1 ppm
sulfur. Unfortunately, the olefin-conditioning stability problems described in Section 5 will also
need to be resolved before commercial use of carrier-facilitated membranes in olefin-related
applications can advance any further.
7. NEW OLEFIN CARRIERS
7.1 Permeation Properties of Membranes Containing Alternate Olefin Carrier Salts
Previous studies at MTR have shown that AgBF4 is an effective carrier for facilitated olefin
transport membranes, whereas various other silver salts, such as AgNO3, are not. It is believed
that small anions like NO3- bind tightly with Ag+, preventing the silver cation from complexing
olefins. In contrast, large, bulky BF4- anions are more easily disassociated, allowing silver ions
to interact with olefin molecules. As part of an effort to identify additional effective (and
perhaps more stable) carriers, a number of different salts possessing large anions were tested for
their ability to facilitate olefin transport in membranes. The results of mixed-gas ethylene/ethane
tests for Pebax® 2533 membranes containing 80 wt% of eight different salts are shown in Table
9. The copper and cobalt salts tested were either insoluble in the Pebax® coating solution or
showed no selectivity enhancement compared to a pure Pebax® 2533 membrane (mixture
ethylene/ethane selectivity of less than 2). The silver salts showed better performance. Two
silver salts, AgPF6 and AgSbF6, exhibited mixed-gas ethylene/ethane selectivities of greater than
10. However, these selectivities are still lower than AgBF4 (C2H4/C2H6 ~30) and more
importantly, ethylene permeances in membranes containing 80 wt% AgPF6 or AgSbF6 are
approximately an order of magnitude lower than those in AgBF4–containing membranes. For
these reasons, from a transport standpoint, none of the salts tested are competitive with AgBF4.
59 311 Final report

DE-FC36-04GO14151
Table 9. Mixed-gas Transport Properties of Pebax® 2533 Membranes Containing 80 wt%

of Various Carrier Salts
Mixed-gas Permeance(gpu) Mixed-gas

Carrier Salt
Co(BF4)2 1.4* 1.0* 1.4*
Cu(BF4)2 0.5 0.3 1.5
CuBH4((C6H5)3P)2 Not soluble in casting solution
AgC2F3O2 2.8 1.2 2.6
AgCF3SO3 0.6 0.2 3.0
AgPF6 3.6 0.3 11
AgSbF6 2.4 0.2 14
AgBF4 (standard carrier) 30 1.0 30
Test conditions: feed pressure = 50 psig, T = 23°C. Mixture composition: 65% C2H4, 35% C2H6.
*Measured with pure gases at 50 psig and 23°C.
Another important criterion for evaluating an olefin carrier is its stability. As discussed in
Section 6, membranes formed from Pebax® 2533+AgBF4 are deactivated by light and reducing
gases such as hydrogen, sulfur species, and acetylenes. This deactivation occurs because Ag+
carrier ions are reduced to silver metal or other reaction products that are not able to complex
olefins. Such exposure causes Pebax® 2533+AgBF4 membranes to turn black (due to the in-situ
creation of silver metal particles) and gradually lose their high olefin/paraffin selectivity. An
ideal carrier would exhibit greater stability than AgBF4 while yielding membranes with at least
comparable transport properties.
To see whether another silver salt has better stability than AgBF4, a test was conducted with
Pebax® 2533 membranes, one containing 80 wt% of AgBF4 and another containing 80 wt%
AgPF6 (silver hexafluorophosphate) carriers. Membranes prepared with the different carriers
were exposed to light and tested over a five day period. Both membranes gradually turned
brown, suggesting that Ag+ ions were being reduced. Consistent with this observation, the
ethylene permeances and ethylene/ethane selectivities decreased over time for both carriers. At
the end of the exposure period, AgBF4 membranes maintained significantly better performance
than those prepared with AgPF6. For example, the mixed-gas ethylene permeances of the
membranes after five days were 10 gpu and 1.3 gpu for AgBF4 and AgPF6, respectively. This
light-exposure test suggests that the best performing alternative carriers examined to date are not
substantially more stable than AgBF4. This result is not unexpected as both of these salts utilize
silver ions as the carrier.
7.2 Permeation Properties of Membranes Containing Mixed Carrier Salts
To date, carriers other than silver have shown little promise by themselves. Work with mixed
carriers has shown more potential, however. Table 10 presents data for Pebax® 2533 membranes
containing various mixtures of both copper and silver carriers. Cu(I) ions have been reported in
the literature to reversibly complex olefins and to resist acetylene poisoning. Because we had
difficulty dissolving a Cu(I) salt in our polymer/solvent system (Pebax® 2533/ethanol),
Cu(BF4)2, a Cu(II) salt, was used. As shown in Table 3, Cu(BF4)2 by itself was not effective at
facilitating olefin transport. However, when blended with AgBF4, it yielded membranes with
60 311 Final report

DE-FC36-04GO14151
better performance than those containing the silver salt alone. For example, although 40 wt%
AgBF4 membranes exhibited a mixed-gas C2H4/C2H6 selectivity of only 6, membranes
containing 40 wt% AgBF4 and 40 wt% Cu(BF4)2 showed a mixture selectivity of 15. This
performance enhancement is not specific to copper salts (or even metal salts), as demonstrated
by the results with NH4BF4, also shown in Table 3. Pebax® 2533 membranes containing 40 wt%
AgBF4 and 40 wt% NH4BF4 showed a mixed-gas C2H4/C2H6 selectivity of 20. These findings
suggest that the BF4- anion may play a more significant role in transport than previously
believed.
Table 10. Mixed-gas Transport Properties of Pebax 2533 Membranes Containing Mixed
Carrier Salts.
Mixed-gas Permeance (gpu) Mixed-gas C2H4/C2H6

Carrier
C2H4 C2H6 selectivity
40 wt% AgBF4 3.9 0.7 6.0
80 wt% Cu(BF4)2 0.5 0.3 1.5
20 wt% AgBF4 +
1.4 0.2 6.0
60 wt% Cu(BF4)2
40 wt% AgBF4 +
4.4 0.3 15
40 wt% Cu(BF4)2
60 wt% AgBF4 +
6.0 0.2 29
20 wt% Cu(BF4)2
40 wt% AgBF4 +
10 0.5 20
40 wt% NH4BF4
60 wt% AgBF4 +
20 0.6 32
20 wt% NH4BF4
®
Membrane configuration: 1 x 0.5% Teflon AF2400 / 2 x 2% Pebax 2533 + salt / PVDF. Test conditions: feed pressure
= 50 psig, T = 23°C. Mixture composition: 65% C2H4, 35% C2H6.
To further investigate this mixed carrier discovery, several additional salts were examined. The
results of mixed-gas tests are shown in Table 11. Consistent with our earlier measurements,
some of these non-silver salts also improve performance when added to Pebax® 2533 + AgBF4
membranes. For example, the membrane containing 40 wt% Ca(BF4)2 exhibited a mixed-gas
ethylene/ethane selectivity of 27, nearly equivalent to a standard 80 wt% AgBF4 membrane.
Another interesting finding is that it appears the non-silver salt does not have to contain a BF4-
anion to enhance membrane performance. One of the membranes containing Mg(CF3SO3)2 also
shows enhanced performance compared to a membrane with AgBF4 alone. Generally speaking,
this mixed carrier discovery offers the possibility to obtain better membrane performance using
significantly less of the expensive silver salt.
61 311 Final report

DE-FC36-04GO14151
Table 11. Mixed-Gas Transport Properties of Additional Pebax® 2533 Membranes

Containing Mixed Carrier Salts
Mixed-gas Permeance (gpu) Mixed-gas C2H4/C2H6

Carrier
Ethylene Ethane Selectivity
40 wt% AgBF4 3.9 0.7 6
40 wt% AgBF4 +
40 wt% NaBF4 9.1 1.0 9.1
40 wt% AgBF4 +
40 wt% NH4BF4 7.0 0.6 12
40 wt% AgBF4 +
40 wt% Fe(BF4)2 2.6 .073 3.5
40 wt% AgBF4 +
40 wt% Mg(CF3SO3)2 5.1 4.7 1.1
60 wt% AgBF4 +
20 wt% Mg(CF3SO3)2 14 0.68 21
40 wt% AgBF4 +
40 wt% Zn(BF4)2 7.4 0.4 19
40 wt% AgBF4 +
40 wt% Ca(BF4)2 23 0.80 27
60 wt% AgBF4 +
20 wt% Ca(BF4)2 26 0.87 30
40 wt% AgBF4 +
40 wt% Al(CF3SO3)3 43 43 1.0
Feed gas = 65% ethylene, 35% ethane: feed pressure = 50 psig; permeate pressure = 0 psig at room temperature.
The reason for the improved performance of mixed carrier systems is not well understood.
Previously, it has been noted that metal salts can increase olefin sorption in aqueous silver salt
solutions. It was suggested that the metal cations attract BF4- and water molecules, leaving Ag+
ions more exposed, which allows olefin molecules to interact with silver more easily. Based on
this hypothesis, the charge-to-size ratio of the non-silver metal ion should be as high as possible.
Comparing the data for mixed carriers with 40 wt% AgBF4+40 wt% non-silver salt, the smallest
divalent borofluorate salt, Ca(BF4)2, gives the highest ethylene permeance and selectivity. This
result suggests that smaller salts, such as Mg(BF4)2 or Be(BF4)2, might enhance silver salt
membrane performance even further.
An additional benefit of the mixed carrier systems is that they appear to degrade at a slower rate
when exposed to light than membranes containing only AgBF4. Because such behavior suggests
a possible increase in membrane stability, this phenomenon was further investigated. Various
coating solutions and membrane samples were prepared and images recorded. Figure 43(a)
shows the membranes on the day they were made, while Figure 43(b) shows the membranes six
days later. Our standard 80 wt% AgBF4 membrane (shown on the left) appears dark brown in
color almost immediately after coating. The dark brown color signifies the presence of silver
metal particles, indicating that active Ag+ ions have been reduced to inert Ag0. Based on Figure
43, membrane samples containing Ca(BF4)2 or Zn(BF4)2 do not exhibit immediate silver carrier
reduction. This suggests these salts have some potential capacity to stabilize Pebax®
2533+AgBF4 membranes. Although not shown, this was not the case for NH4BF4-containing
membranes, which rapidly turned black, much like the standard AgBF4 membrane. In contrast,
62 311 Final report

DE-FC36-04GO14151
both Al(CF3SO3)3 and Mg(CF3SO3)2 also showed some capacity to mitigate light-induced silver
reduction.
As discussed in Section 5, we found that adding an H2O2/HBF4 mixture to our standard Pebax®
2533 + AgBF4 membrane retarded Ag+ ion reduction in freshly coated membranes. However,
because H2O2 and HBF4 are volatile liquids that evaporate over time, their beneficial effect was
short-lived unless the membranes were stored in a sealed environment. Ca(BF4)2 and Zn(BF4)2
are solids, and thus not subject to this evaporative limitation. Consequently, they have greater
potential as long-term membrane stabilizers. Figure 43 demonstrates this point, as these mixed
carrier membranes, especially the one containing Ca(BF4)2, appear white after six days of light
exposure. Over this time, our H2O2/HBF4-containing membranes would begin to turn black,
indicating the presence of reduced silver.
(a) (b)
Figure 43. A comparison of membranes containing various mixed carrier salts on (a) the first
day they were made and (b) six days after they were prepared. In both pictures,
the composite Pebax® membrane on the left contains 80 wt% AgBF, the one in
the middle contains 40 wt% AgBF4+40 wt% Ca(BF4)2, and the membrane on the
right contains 40 wt% AgBF4+40 wt% Zn(BF4)2. The corresponding solutions are
shown above the membranes.
Although time did not permit a thorough aging study of the mixed carrier membranes, Figure 44
presents limited data for the most promising mixed carrier, Ca(BF4)2. Initially, the ethylene
permeances of an 80 wt% AgBF4 membrane (the control) and a 60 wt% AgBF4 + 20 wt%
Ca(BF4)2 are approximately equal, while that of a 40 wt% AgBF4+40 wt% Ca(BF4)2 membrane
is about half as high. Over time, both of the higher permeance membranes show a decline in
performance, while the ethylene permeance of the 40 wt% AgBF4+40 wt% Ca(BF4)2 membrane
appears to be stable. However, the mixed-gas ethylene/ethane selectivities of all three
membranes, which are initially between 25-30, decline over the course of the experiment. This
decrease is most rapid for the 40 wt% AgBF4+40 wt% Ca(BF4)2 membrane. Interestingly, the
reason selectivity decreases for this membrane is because ethane permeance increases, rather
than that the permeance of the facilitated gas (ethylene) decreases. This would suggest that the
63 311 Final report

DE-FC36-04GO14151
membrane ionic character is decreasing over time, which is somewhat different than the olefin-
conditioning effect that impacts high silver salt content membranes. These mixed carrier
membranes clearly warrant further investigation.
22 35
(a) (b)
Mixed-gas C2H4 permeance (gpu)
Mixed-gas C2H4/ C2H6 selectivity

20 30 20 wt% Ca(BF4)2
18 25
80 wt% AgBF4
16 20 80 wt% AgBF4
14 15
20 wt% Ca(BF4)2
40 wt% Ca(BF4)2
12 10
40 wt% Ca(BF4)2
10 5
0 1 2 3 4 5 0 1 2 3 4 5
Figure 44. Mixed-gas (a) ethylene permeance and (b) ethylene/ethane selectivity as a
function of permeation time for Pebax® 2533 composite membranes containing 80
wt% AgBF4, 60 wt% AgBF4 + 20 wt% Ca(BF4)2, and 40 wt% AgBF4 + 40 wt%
Ca(BF4)2.
7.3 Development of Non-Silver Salt Olefin Carriers
The industrial separation of olefins from paraffins using complexation agents has not yet come to
fruition on an industrial scale. The perceived reasons for this are (1) a potential threat of
explosion hazard with the current best-performing complexation agent - silver - or, less
preferably, copper, and (2) deactivation of silver or copper complexing agents by potential
impurities in the feed, particularly acetylene, hydrogen, and hydrogen sulfide (as described in
Section 6). In this project, we began the work required to address these issues, using novel
complexing agents with 100% selectivity for olefins over paraffins. Dr. Marianna Asaro of SRI
International (SRI) led this research.
SRI performed the research in two tasks. The first task was to comprehensively review the
literature on olefin complexation. Hundreds of olefin complexes exist, but almost none of them
have been used to address this application. Hence most of the data of interest to make selections
were unavailable. We therefore used what information was available to prioritize candidates and
then undertook the second task, which was the synthesis of candidate complexes. The
complexes were then introduced into MTR’s membrane materials and the amount of ethylene
uptake was compared to that obtained using only the membrane without any complexing agent
64 311 Final report

DE-FC36-04GO14151
and to that obtained using the silver-loaded membrane system. The strategies employed in these
Tasks are described below.
Task 1. Identification of Candidate Complexing Agents
Olefin complexes or olefin binding chemistries were identified through searching the literature
and then reading the primary references. The complexation agents were evaluated according to
the following criteria; the criteria are in approximate order of importance.
a. Reversibility of Olefin Binding
The potential efficacy of olefin uptake in a separation system depends fundamentally on the
equilibrium constant of olefin binding, KBIND, and the rate constants of olefin binding, kfor, and
olefin dissociation, krev. The relationship between these parameters is shown in Equation (1):
kfor
KBIND =
krev
What the values of these parameters should be for a good olefin carrier in a solid polymer
electrolyte membrane is not clearly established. The optimal equilibrium binding constant,
KBIND, for a liquid phase separation system has been estimated by Kemena et al as ~ 10. The
optimal complexation agent for a solid phase system, as in the present work, may be somewhat
different than that for a solution phase system; however, the solution phase precedent was taken
as a starting approximation for the solid phase work.
In principle, a complex having KBIND of 10 would release 90% of its bound olefin with a pressure
differential across the membrane of about 10 and 99% of its bound olefin with a pressure
differential of about 100. Larger pressure differentials than these become uneconomical. Lower
binding constants than ~ 10 can be used, with the caveat that as KBIND approaches 0, the
selectivity for olefin over paraffin imparted by the complexation agent also approaches zero.
The equilibrium binding constant KBIND is a ratio of rates of the forward reaction (olefin binding)
and the reverse reaction (olefin dissociation). The values of these kinetic constants are as
important as their ratio. Both kfor and krev need to be large, with the former being ~ 10 times
larger than the latter. The required magnitudes of these rate constants are highly specific to the
type of separation system. The observed rate constant, kobs, for any reaction is particularly
affected by the medium and may differ substantially with different separator configurations (for
example, membrane, packed bed, aqueous solution, etc.) due to differences in mass transfer
characteristics or limitations.
Those few non-silver based systems that have been characterized for equilibrium binding of
olefins have been studied in solution phase; in fact, most literature describing olefin complexes
or complexation does not provide any data on equilibrium constant or rate constants of olefin
binding. Nonetheless, we were able to down-select and prioritize by careful reading and
deductive reasoning. For example, an olefin complex that was synthesized and then dried under
a typical vacuum of 1 torr or below without change in composition is likely a poor choice
65 311 Final report

DE-FC36-04GO14151
because KBIND is presumably >> 10 or the olefin would have been released. It is also possible to
glean qualitative or quantitative information on olefin binding from spectroscopic results,
particularly from the observation of olefin exchange by NMR spectroscopy. Such solution phase
data are useful as guidelines to down-selection.
Complexes that display reversible olefin binding must do so by the appropriate mechanism to be
useful for gas separations. Many olefin complexes exchange olefin associatively, only losing
olefin after taking up another molecule such as solvent. Those that exchange or lose olefin
dissociatively are much more attractive for reversible gas uptake applications.
In summary, the potential of a candidate complexing agent to display reversible olefin uptake
was assessed by consideration of its equilibrium binding constant, kinetic constants, and reaction
mechanism where such data were available or could be qualitatively assessed. Those systems
where no such data were found could still be good candidates but were given lower priority in
favor of those for which positive information was available.
b. Stability
The thermal and chemical stability of the candidate complexing agents were also important
selection criteria in this work. The thermodynamics of olefin binding were considered first, as
described under ‘Reversibility of Olefin Binding’. The thermal stability of the complexes toward
other decomposition routes was considered next in the Task 1 screening; for example, when an
olefin complex decomposes through some route other than loss of olefin.
The stability of complexes toward acetylene, other olefins such as propylene, hydrogen sulfide,
acetylene, hydrogen, and oxygen were also canvassed in the literature screening. Typically, little
if any of this information was available. Instability in the presence of oxygen, if present, is often
noted in the literature, and occasionally competitive uptake experiments with ethylene and an
acetylenic compound are provided. Stability toward hydrogen sulfide is almost never reported;
however, some sense of this may be obtained by noting the stability toward water. Hence, down-
selection based on stability toward potential contaminants was made primarily based on SRI’s
two decades of previous experience with these types of complexing agents. In general, chemical
stability is most likely to exist where chemical characteristics are relatively neutral. Thus the
candidate complexes will likely be neither highly basic nor highly acidic and neither highly
oxidizing nor highly reducing in character.
c. Polarity and Solubility
The next screening consideration was the polarity and solubility of the complexing agent, as it
would affect compatibility with the MTR membrane systems. The ideal complexing agents
would be soluble in water, alcohols or other polar solvents in which MTR’s preferred membrane
materials could be dissolved. PEO-based membranes rather than Pebax®-based membranes were
used for testing new olefin carriers because of their water solubility and easier analytical testing
properties. The selection was not entirely limited to the more polar species, however, because
MTR has a repertoire of other membrane polymers. An additional advantage of using polar
solvents is that ethane has low solubility in polar environments.
66 311 Final report

DE-FC36-04GO14151
Qualitative solubility information is often available in the literature. The preferred candidates
based on solubility tended to be relatively polar, neutral species or cationic species with
solubilizing counterions. Anionic candidates were few in number and were generally considered
too chemically or thermally unstable for use in this application.
d. Ease and Cost of Synthesis
Ease of chemical synthesis was the final consideration in the screening process, albeit with less
weight than the other criteria at this stage of development. The few commercially available
complexes that bind olefins were considered and tested as appropriate. Most candidates would
need to be synthesized, however, and only those that were deemed definitely unscalable (for
plant use at industrial production levels) were rejected based on synthesis method. Likewise, the
projected cost of the complexing agent was given low screening weight, because in our
experience an expensive, but effective and long-lived complexing agent or catalyst is well-
tolerated in the process economics. For example, the catalyst used for commercial preparation of
the commodity chemical ethylene oxide can be as much as 35% silver by weight, but the
material is sufficiently robust to provide favorable process economics.
Task 2. Synthesis of Candidate Complexing Agents
The candidate complexes were synthesized according to literature methods. The identity and
purity of complexes were confirmed by
• NMR spectroscopy of organic segments of the complexing agents, including bound and
free ethylene;
• IR spectroscopic analysis of various groups, particularly the nC=C stretch of the bound
ethylene;
• UV/visible spectroscopy;
• observation of changes in absorption position depending on whether the ethylene was
bound or unbound; and
• comparison to literature data for proof of identity.
Discussion of Screening and Synthesis Results
Approximately 250 olefin complexes or olefin binding chemistries were identified through
literature search, reading the primary references in the initial literature search and perusing other
references. These were grouped according to the categories shown in Table 12; the categories
are numbered in order of Task 1 screening priority. Approximately 40 complexing agents were
deemed of high interest, either based on data found in the literature or based on our experience.
67 311 Final report

DE-FC36-04GO14151
Table 12. Categories of Olefin Complexes Considered for Olefin/Paraffin Separation.

Categories are numbered in order of Task 1 screening priority.
Priority Category Description

1 Cationic Olefin Complexes of Ru
1a Cationic Olefin Aqua Complexes of Ru
1b [Ru(amine)n(olefin)]2
2 Neutral Olefin Complexes of Ru
3 Cationic Complexes of Rh/Ir/Co with Olefins
3a Cationic Olefin Aqua Complexes of Rh
3b Cationic Bis(olefin) Complexes of Rh
4 Neutral Olefin Complexes of Rh/Ir
4a Neutral Bis(Olefin) Complexes of Rh/Ir
5 Olefin Complexes of Fe
6 In-Situ Formation of Cationic CO Olefin Complexes of Groups 6 and 7
7 Neutral Carbonyl Olefin Complexes of Groups 6 and 7
7a In-Situ Formation of Neutral CO Olefin Complexes of Groups 6 and 7
8 Group 6 Olefin Complexes with Oxy Ligands
9 Metallocenes with Olefin Ligands
10 Olefin Complexes of Co
11 Olefin Complexes of Ni
12 Olefin Complexes of Pt and Pd
13 Reversible Metal-Hydrogen (M–H) Insertion
14 Dithiolate Complexes
15 Metallocubanes
16 Complexes of Olefins & Lewis Acids
17 Special Reagents
18 Complexes of Olefins with Alkali Metals and Transition Metals
19 Complexation of Olefins with Main Group Elements
20 Olefin Complexes of Ag, Cu, Au, and Hg
21 Patents
Project time and budget constraints considerably limited our experimental synthesis and testing
efforts beyond the initial screening exercise. Consequently, priority was given to complexes for
which some data were available. A total of eight complexes were synthesized and two additional
complexing agents were purchased. Of these, four were selected for testing in MTR’s membrane
system. Some of the complexes that showed promise based on the membrane tests are the
subject of an SRI patent application that is still in preparation, so the complexes are referred to
below by their identification numbers. The discussion of each complex provides an example of
the variation in information, synthesis, and results that were obtained in this section of our
project.
a. Complex 14646-16 was selected for evaluation because it has good stability and was
anticipated to have good compatibility with polyether membranes of interest for olefin/paraffin
gas separations. Because this olefin complex is cationic, most of its salts will tend to have
solubility in the polar solvents needed to dissolve the polymer and to cast the membrane. The
ethylene complex of 14646-16 has been reported in the literature without equilibrium or rate
constant data. The identity of the recrystallized ethylene complex was established by IR
68 311 Final report

DE-FC36-04GO14151
spectroscopy (nC=C at 1537 cm-1, weak, Nujol mull) and by NMR spectroscopy (d = 3.18 ppm in
D2O, singlet, bound ethylene), and no side products were observed.
Complex 14646-16 is known to bind acetylene, forming a 1:1 adduct without formation of any
acetylides or acetylene polymers. Based on a comprehensive investigation of the literature, we
find that it is not feasible to predict, a priori, whether acetylene or ethylene binds more strongly
(if at all) to a metal complex in a given environment. For candidates such as complex 14646-16,
if ethylene binding experiments showed promise, we would then perform the ethylene/acetylene
competition experiments within the membrane.
The BF4 salt of complex 14646-16 was selected for synthesis because it had good solubility and
by analogy to the preferred use of BF4 in the silver system, where we believe that BF4 exerts a
synergistic effect with the complexation agent in olefin/paraffin separations.
The ethylene-bound form of complex 14646-16-BF4 was co-dissolved in 24 ml of water at 2

wt% each of the complex and the PEO polymer. The membrane was cast and the solvent was
evaporated at ambient temperature over several days. The initially light yellow, compound-
loaded membrane slowly changed color to purple over this time period and the dried membrane
did not show desorption of ethylene. We then prepared the PF6 salt, 14646-16-PF6, and
performed the casting in acetonitrile, with solvent removal over a few hours under reduced
pressure. The resulting light yellow, loaded membrane was stable but did not show ethylene
desorption. The fact that neither complex 14646-16 salt showed ethylene desorption suggests
that KBIND for this complex is too high for reversibility of complexation.
b. Complex 14646-37 was selected because it has good stability and is also commercially
available. It is a neutral species that is not soluble in the polar solvents needed for PEO
membranes, so we dissolved the ethylene-bound form of 14646-37 at 2 wt% contained metal in
acetone and used ethylcellulose at 2 wt% to cast the membrane. The dried membrane formed a
powder instead of a film, which suggests that the complex may have interacted with the polymer.
The commercial ethylcellulose was 47% ethoxy-substituted, the rest being hydroxy-substituted.
We suspect that these hydroxyls on the polymer bonded to the carrier complex and destroyed the
film-forming ability of the system. We were unable to obtain sorption data on the powder.
c. Complex 14646-42 was selected because it has good stability and its direct precursor is
commercially available. It is a neutral species that is not soluble in the polar solvents needed for
PEO membranes, so we co-dissolved the unbound complex and polystyrene at 2 wt% each in a
solution of 25 mL acetone and 5 mL toluene. The membrane was dried but failed to show
ethylene uptake. This was surprising; however, prior work in our labs with the ethylene complex
of 14646-42 was performed in ethanol/water solution phase and apparently that work does not
transfer to the polystyrene membrane system.
d. Complex 14646-48 was selected because it has good stability and has an estimated solution
phase equilibrium dissociation constant of about 0.04 M-1. This corresponds to a KBIND of about
25 M-1, which is a bit larger than our target range of about 10 + 5 M-1 but might be sufficient for
proof-of-principle. Complex 14646-48 was prepared in four steps from commercially available
starting materials. The product was recrystallized and characterized for identity and purity by
69 311 Final report

DE-FC36-04GO14151
proton NMR spectroscopy. Stability tests were not performed but we did note that the yellow
solid was stable to air during handling.
A quantity of the ethylene-bound form of 14646-48 corresponding to 1.4 mmol of ethylene-

binding capacity was co-dissolved with PEO polymer at 1.2% : 1.2% w/w concentration of
contained complexation metal : PEO polymer. Acetonitrile was used as solvent. The dissolution
was completed by allowing the mixture to agitate on a roller mill overnight. The compound
dispersed well and formed a good film after casting and evaporation of the solvent. Initial
studies showed no ethylene desorption. The sample was kept under reduced pressure for some
days, whereupon loss of ethylene had occurred. The sample was cycled with ethylene
atmosphere and showed re-uptake of ethylene (see Figure 41). These results suggest that the
complex is able to reversibly uptake olefin and that a modified version of this complex having a
somewhat lower value of KBIND should be designed, prepared and tested. That work is fairly
straightforward to do; however, the project ended before it could be performed.
Figure 45 shows an ethylene sorption isotherm for the PEO/complex 14646-48 blend. For
comparison, ethylene sorption isotherms for pure PEO, PEO containing two compounds
previously examined in this program, and PEO containing AgBF4 are also shown. In contrast to
previous complexes examined, 14646-48 appears to increase ethylene sorption when dispersed in
PEO. This is an encouraging finding, although the amount of ethylene uptake is still more than
an order of magnitude lower than that associated with PEO/AgBF4 blends.
10
PEO + 50 wt% AgBF4

divided by 10
8
Ethylene 6
concentration
( )
cm 3(STP) PEO + 50 wt% PEO + 45 wt%
complex complex 14646-48
cm 3 polymer 4 14646-16-PF 6
PEO + 50 wt%
complex
14646-16-BF 4
2
PEO
0
0 1 2 3 4 5 6
Pressure (atm)
Figure 45. Ethylene sorption isotherms for PEO and PEO containing various olefin
complexing agents. The ethylene uptake data for PEO + 50 wt% AgBF4 have
been divided by 10 for presentation purposes.
70 311 Final report

DE-FC36-04GO14151
To test the feasibility of using complex 14646-48 as an olefin carrier in a facilitated transport
membrane, we prepared PEO composite membranes containing 80 wt% of 14646-48. These
membranes showed no selectivity for ethylene over ethane in pure-gas tests. It appears that the
ethylene capacity of 14646-48 is not high enough or the binding constant is too great for
effective use in a solid polymer electrolyte membrane.
e. Complex 14646-91 was selected because it has good stability and is of a similar structural type
to 14646-48; however, we determined that it has a value of KBIND of about 800 M-1 and so its
ethylene complex is too strongly bound and is not a good candidate for this application.
Based on the number of issues identified in this project that would need resolving in order to use
silver salts for olefin/paraffin membrane separations, the discovery of a new, stable olefin carrier
could be the key breakthrough to allow commercialization of carrier-facilitated membranes.
Certain of the complexes tested show good potential for successful development, and deserve
further study. This would allow strengthening of the proof-of-principle determined during this
project.
8. OLEFIN/PARAFFIN SEPARATION APPLICATIONS
8.1 Introduction
At the start of this program three target applications were identified as potential areas where
facilitated transport membranes could be used for olefin/paraffin separations. The three initial
target applications are briefly described below.
Steam Cracker Cold Trains. Steam crackers produce a mixture of hydrogen, carbon dioxide,
light olefins, and paraffins. After treatment to remove carbon dioxide and other impurities, the
gas is compressed to 500-600 psi and cooled, liquefied and separated in a series of large towers.
This cold train represents the bulk of the capital and operating costs of the plant. We initially
proposed using a membrane separation step to treat the gas produced by the steam cracker. The
membrane unit was expected to cost approximately $30-40 million. However, the associated
cracker cold train would be significantly smaller, producing capital cost savings and very large
energy savings.
Reactor Vent Streams. Technology for the recovery of propylene and ethylene from the paraffin
vent streams of petrochemical reactors has the potential for widespread use. We estimate that as
many as 200 U.S. petrochemical plants could adopt this technology. An additional 300-400
potential users are located outside the United States. Individually, these applications are small;
taken together, they represent a significant opportunity for resource recovery. This gas is usually
from a noble metal catalytic reactor; sulfur and acetylene levels are low. This application would
also provide the opportunity to introduce and commercialize olefin/paraffin membrane
separation technology on a small scale with favorable economics before trying to overcome the
barriers involved in building the first membrane-based world-scale steam cracker gas separation
plant.
71 311 Final report

DE-FC36-04GO14151
FCC Off-gas. Most U.S. and European refineries use fluid catalytic cracker (FCC) units to
convert heavy oil fractions to lower-molecular-weight hydrocarbons for the gasoline pool. In this
process, a fraction of the feed is converted to hydrogen, methane, ethylene, propylene and
paraffins. The hydrogen and olefins in this gas have considerably greater value when separated
than when used as refinery fuel. A few refineries have installed turbo expander-distillation units
to treat this gas, but most U.S. refineries continue to use this gas as fuel due to the high cost of
liquefaction and distillation. Olefin/paraffin membrane separation technology could potentially
expand the number of refineries capable of economically separating FCC off-gases.
8.2 Olefin Metathesis
During this project, a promising application requiring an olefin/paraffin separation technology

was identified by our ABB Lummus partners. In this process, the streams of interest are n-
butene/n-butane mixtures. Developing an economical method of separating these mixtures is
important in olefin metathesis, used to produce polymer-grade propylene by combining the n-
butene with ethylene. This separation is not commercially attractive using distillation. If
successful, calculations show that a membrane process would increase reactant utilization of
metathesis plants by 6%. The energy savings achieved by producing 6% more propylene from
the same volume of feedstock are very large.
While butene/butane separations are required in metathesis operations, this separation is also
important in the production of 1-butene for use as a polyethylene comonomer and polybutene-1
monomer; and 2-butene for use as a methyl ethyl ketone feedstock.
A process flow diagram of an olefin conversion metathesis plant is shown in Figure 46.[40]
Some plants are built as stand-alone units, but many are integrated with an existing steam cracker
that produces the ethylene and butene feedstocks. In a metathesis unit, 0.74 tons of butenes,
worth approximately $200/ton are reacted with 0.31 tons of ethylene worth approximately
$500/ton to produce one ton of propylene valued at $550/ton (2005 prices). As a result, a feed
upgrade of $247/ton is achieved. Propylene demand is growing at 6%/year in the U.S., and even
more rapidly in other regions of the world. As a consequence of this growth, many metathesis
plants are being installed. Total worldwide production of propylene by metathesis is 4.5 million
tons per year — approximately 6% of total propylene production.
72 311 Final report

DE-FC36-04GO14151
Figure 46. Process flow diagram of a metathesis olefin conversion process. Approximately
10% of the n-butene feedstock is lost with the n-butane purge.
In the process shown in Figure 46, ethylene and mixed butenes are heated and sent to a catalytic
metathesis reactor. The catalyst promotes the reaction of ethylene and 2-butenes to produce
propylene, and simultaneously isomerizes 1-butene to 2-butenes. Typical per-pass conversions of
almost 60 to 70% are achieved with about 95% selectivity to propylene. The reactor product gas
is cooled and fractionated to remove ethylene, which is recycled to the reactor feed. A small
portion of the recycled ethylene is purged to remove methane, ethane and other light gas
impurities. The ethylene column bottoms are then fed to the propylene column. There is no
propane present, so it is not necessary to have a super-fractionator column to produce polymer-
grade propylene. The bottoms from the propylene column contain all the unreacted butanes and
are recycled to the front of the metathesis reactor. The n-butene feed to the process contains
approximately 10% n-butane and small amounts of isobutane. This n-butane rapidly builds up in
the recycle loop until the butene mixture entering the reactor contains as much as 25% butane.
This high level of butane requires energy to heat the gas and lowers the production rate of the
reactor. n-Butane levels are controlled by purging a portion of the butene recycle stream. This is
a large purge stream and approximately 0.7 mols of n-butene are lost with every mole of n-
butane purged (~ 2,000 kg/h of n-butene in a world-scale metathesis plant). The objective of our
membrane process is to reduce the n-butene loss to 0.1 mol of n-butene/mol of n-butane purged
(~ 300 kg/h).
Figure 47 shows the performance of Pebax® 2533+AgBF4 membranes with a 50/50 n-butene/
isobutane gas mixture as a function of temperature. The feed in these experiments was a 50/50
vol% C4 mixture near saturation pressure (activity >0.95). The permeances for the olefin, 1-
butene, are very high (>400 gpu) and decrease with increasing temperature. This temperature
effect is typical for membranes where permeation is solubility controlled (rather than diffusion
controlled as in size-sieving membranes). The extremely high permeance for 1-butene is likely
73 311 Final report

DE-FC36-04GO14151
related to extensive membrane swelling caused by high sorption of this large penetrant. Even in
this swollen condition the membrane still exhibits good selectivity for 1-butene over isobutane
(about 8).
1,000 10
Mixed-gas 1-butene/isobutane selectivity

(a) (b)
9
1-butene
6
100
isobutane 5
3
35 40 45 50 55 60 65 35 40 45 50 55 60 65
Temperature (°C) Temperature (°C)
Figure 47. Mixed-gas 1-butene/isobutane (a) permeance and (b) selectivity as a function of
temperature.
If stable facilitated membranes with the properties shown in Figure 47 were realized, the
required membrane area would drop by nearly an order of magnitude, making the n-butene/n-
butane membrane separation process that much more attractive. ABB Lummus estimates
membrane recovery systems for olefin metathesis at approximately a $150 million opportunity.
If the membrane technology could be successfully developed for this first particularly attractive
application, other related applications could be developed for upgrading butenes in refinery and
other petrochemical operations.
In light of the facilitated membrane stability issue described throughout this report, it seems
unlikely that this technology will be ready in the near future. Consequently, we are working on
developing passive membranes that could be used for the metathesis olefin/paraffin application.
9. CONCLUSIONS
In this project, we continued the development of a new class of facilitated olefin transport
membranes identified in a previous NSF project at MTR. These solid polymer electrolyte
membranes initially showed promisingly high permeances and olefin/paraffin selectivities even
in the absence of water. The focus of the research in this project was a detailed analysis of
membrane time-dependent performance, including performance after exposure to carrier
poisoning species.
74 311 Final report

DE-FC36-04GO14151
We discovered that silver salt-based facilitated transport membranes are not stable even in the
presence of ideal olefin/paraffin mixtures. For example, over a five-day period of continuous
operation, the mixed-gas ethylene permeance of Pebax® 2533+80 wt% AgBF4 membranes
decreases from 50 to less than 1 gpu. Over the same time frame, the ethylene/ethane selectivity
decreases from 40 to less than 2. This membrane performance decline appears to be caused by a
previously unrecognized phenomenon that we have named olefin conditioning. As the name
implies, this mechanism of performance degradation becomes operative once a membrane starts
permeating olefins. Prior to olefin exposure, silver-salt membranes exhibit completely stable
performance when exposed to inert (non-facilitated or non-poisoning) gases. Olefin conditioning
is distinct from carrier poisoning in that it does not result in reduction of the silver carrier ion. In
fact, visual inspection and FTIR analysis confirm that active silver ions are still present in olefin-
conditioned membranes. Also, in contrast to carrier poisoning, olefin conditioning causes a
decrease in the permeance of inert gases co-permeating with olefins.
Experiments conducted during this project have shown that olefin conditioning is accelerated by:
• Higher membrane silver salt content – for dry Pebax® 2533 membranes with less than 40
wt% AgBF4, performance is stable over a one-week period. However, olefin/paraffin
selectivities are not useful at these low silver loadings.
• Higher olefin feed pressure – at a fixed feed pressure, the rate of membrane performance
decline increases with increasing olefin feed concentration.
• Higher temperature – at 5°C, membrane olefin permeance and olefin/paraffin selectivity
appear to be relatively stable. The rate of performance degradation increases
monotonically with increasing temperature.
• The presence of organic plasticizers or water – low-molecular-weight polyethylene oxide
and water are plasticizers that swell the membrane and increase matrix mobility. Both of
these additives enhance the rate of performance decline.
These results suggest that the structure of facilitated transport membranes changes over time due
to olefin conditioning. When the membrane is exposed to an olefin, the polymer matrix becomes
highly swollen as olefin sorbs into the membrane to react with silver ions. In this mobile
environment, structural rearrangement that lowers gas permeance appears to occur within the
membrane – perhaps by clustering of silver ions. This effect is independent of the matrix
polymer (rubbery or glassy) or silver salt (AgBF4, AgPF6, or mixed carriers).
This project is the first study to identify olefin conditioning as a significant factor impacting the
performance of facilitated olefin transport membranes. To date, we have not identified an
effective strategy to mitigate the impact of olefin conditioning. other than running at low
temperatures or with low olefin feed pressures. In our opinion, this issue must be addressed
before further development of facilitated olefin transport membranes can proceed.
In addition to olefin conditioning, traditional carrier poisoning challenges must also be

overcome. Light, hydrogen, hydrogen sulfide, and acetylene exposure adversely affect
membrane performance through unwanted reaction with silver ions. For example, exposure to
pure hydrogen at 35 psia caused reduction of silver ions to silver metal, resulting in an order-of-
magnitude decrease in mixed-gas ethylene/ethane selectivity over a one-week period. Similar
75 311 Final report

DE-FC36-04GO14151
results are obtained for exposure to 10 ppm hydrogen sulfide or acetylene. These harsh
poisoning tests represent a worst-case exposure scenario but nonetheless demonstrate a need to
improve the stability of the olefin complexing agent to develop membranes with lifetimes
satisfactory for commercial application.
A successful effort to improve membrane coating solution stability resulted in the finding that
membrane performance loss could be reversed for all poisoning cases except hydrogen sulfide
exposure. A novel in-situ regeneration treatment with a peroxide/acid mixture in liquid or vapor
form restored aged and poisoned membranes to nearly their initial performance values. This
discovery offers the potential to extend membrane lifetime through cyclic regeneration. We also
found that certain mixed carriers exhibited greater stability in reducing environments than
exhibited by AgBF4 alone. These results offer promise that solutions to deal with carrier
poisoning are possible.
Efforts to identify a novel, non-silver olefin carrier met with some limited success. The best new
carrier tested in this project reversibly sorbed ethylene, but at an insufficient level to enhance
membrane olefin/paraffin selectivity. It is a significant challenge to find complexing agents that
reversibly bind olefins while dispersed in a polymer matrix. However, the payoff for doing so
would be large. As discussed above, carrier stability is still the primary technical hurdle
blocking further development of silver salt-based facilitated transport membranes. If the stability
problems could be largely circumvented by development of a new carrier, it would provide a
clear breakthrough toward finally recognizing the potential of facilitated olefin transport.
During this project a promising application requiring an olefin/paraffin separation technology

was identified by our ABB Lummus partners. In olefin metathesis, a process used to produce
polymer-grade propylene by combining n-butene with ethylene, it is desirable to separate
mixtures of n-butene/n-butane. This separation is not commercially attractive using distillation.
Calculations show that a successful membrane process would increase reactant utilization of
metathesis plants by 6%. The energy savings achieved by producing 6% more propylene from
the same volume of feedstock would be very large. Our facilitated transport membranes show
excellent separation properties for C4 olefin/paraffin mixtures. However, in light of the stability
problems of facilitated membranes, we are also examining high-performance passive membranes
for this application.
The main achievement of this program was the progress made in gaining a more complete
understanding of the membrane stability challenges faced in the use of facilitated olefin transport
membranes. Our systematic study of facilitated olefin transport uncovered the full extent of the
stability challenge, including the first known identification of olefin conditioning and its impact
on membrane development. We believe that significant additional fundamental research is
required before facilitated olefin transport membranes are ready for industrial implementation.
The best-case scenario would be identification of a novel carrier that is intrinsically more stable
than silver ions. Even if such a carrier is identified, development will be required to insure that
the membrane matrix is a benign host for the olefin-carrier complexation reaction and shows
good long-term stability.
76 311 Final report

DE-FC36-04GO14151
REFERENCES
[1] AIChE, "Technology Vision 2020: Separations 200 Road Map," Center for Waste
Reduction Technologies (2000).
[2] R.B. Eldridge, "Olefin/Paraffin Separation Technology: A Review," Industrial &

Engineering Chemistry Research, 32, 2208-12 (1993).
[3] D.J. Safarik and R.B. Eldridge, "Olefin/Paraffin Separations by Reactive Absorption: A
Review," Industrial & Engineering Chemistry Research, 37, 2571-81 (1998).
[4] G.E. Keller, A. Marcinkowsky, S.K. Verma, and K.D. Williamson, "Olefin Recovery and
Purification via Silver Complexation," in Separation and Purification Technology, N.
N. Li and J. M. Carlo, Eds., Marcel Dekker, Inc., New York (1992).
[5] R. Schoellner and U. Mueller, "Influence of Mono- and Bivalent Cations in 4a-Zeolites
on the Adsorption Separation of Ethylene and Propylene from Cracker Gases,"
Adsorption Science and Technolgy, 3, 167 (1986).
[6] I. Pinnau and L.G. Toy, "Solid Polymer Electrolyte Composite Membranes for
Olefin/Paraffin Separation," Journal of Membrane Science, 184, 39-48 (2001).
[7] I. Pinnau, L.G. Toy, and C. Casillas, "Olefin Separation Membrane and Process," U.S.
Patent 5,670,051 (September 23, 1997).
[8] P.F. Scholander, "Oxygen Transport through Hemoglobin Solutions," Science, 131, 585
(1960).
[9] W.J. Ward and W.L. Robb, "Carbon Dioxide-Oxygen Separation: Facilitated Transport
of Carbon Dioxide across a Liquid Film," Science, 156, 1481-84 (1967).
[10] R.D. Hughes and E.F. Steigelmann, "Process," U.S. Patent 3,758,605 (September 11,
1973).
[11] E.F. Steigelmann and R.D. Hughes, "Process for Separation of Unsaturated
Hydrocarbons," U.S. Patent 3,758,603 (September 11, 1973).
[12] R.D. Hughes, J.A. Mahoney, and E.F. Steigelmann, "Olefin Separation by Facilitated
Transport Membranes," in Recent Developments in Separation Science, N. N. Li and J.
M. Calo, Eds., Vol. IX, CRC Press, Inc., Boca Raton, FL (1986).
[13] R.W. Baker, Membrane Technology and Applications, 2nd edition, John Wiley & Sons
Ltd., Chichester, England (2004).
[14] R.D. Hughes and E.F. Steigelmann, "Process," U.S. Patent 3,800,506 (April 2, 1974).
77 311 Final report

DE-FC36-04GO14151
[15] R. Quinn, J.B. Appleby, and G.P. Pez, "New Facilitated Transport Membranes for the
Separation of Carbon Dioxide from Hydrogen and Methane " Journal of Membrane
Science, 104, 139 (1995).
[16] O.H. LeBlanc, W.J. Ward, S.C. Matson, and S.G. Kimura, "Facilitated Transport in Ion
Exchange Membranes," Journal of Membrane Science, 6, 339-43 (1980).
[17] O.I. Eriksen, E. Aksnes, and I.M. Dahl, "Facilitated Transport of Ethene through Nafion
Membranes. Part I. Water Swollen Membranes " Journal of Membrane Science, 85 (1),
89-97 (1993).
[18] S.G. Kimura, W.J. Ward, and S.L. Matson, " Facilitated Separation of a Select Gas
through an Ion Exchange Membrane," U.S. Patent 4,318,714 (March 9, 1982).
[19] C.A. Koval and T. Spontarelli, "Condensed Phase Facilitated Transport of Olefins
through an Ion Exchange Membrane," Journal of the American Chemical Society, 110,
293-95 (1988).
[20] J.D. Way, R.D. Noble, D.L. Reed, G.M. Ginley, and L.A. Jarr, "Facilitated Transport of
CO2 in Ion-Exchange Membranes," AIChE Journal, 33, 480-87 (1987).
[21] W.S. Ho, "Polymeric Membrane and Process for Separation of Aliphatically Unsaturated
Hydrocarbons," U.S. Patent 5,062,866 (November 5, 1991).
[22] W.S. Ho and D.C. Dalrymple, "Facilitated Transport of Olefins in Ag+-containing

Polymer Membranes," Journal of Membrane Science, 91, 13 (1994).
[23] K.V. Peinemann, S.K. Shukla, and M. Schossig, "Permeation and Properties of Highly
Selective Inorganic/Organic Blend Membranes for Separation of Reactive Gases,"
ICOM, pp. 792-94, Chicago, IL (1990).
[24] Y.S. Kang, J.O. Won, B.S. Jung, H.H. Park, and S.G. Oh, "Solid State Polymer
Electrolyte Facilitated Transport Membranes Containing Surfactants," U.S. Patent
6,645,276 (November 11, 2003).
[25] J.H. Kim, B.R. Min, C.K. Kim, J. Won, and Y.S. Kang, "New Insight into the
Coordination Mode of Silver Ions Dissolved in Poly(2-ethyl-2-oxazoline) and its Relation
to Facilitated Olefin Transport," Macromolecules, 35, 5250-55 (2002).
[26] J.H. Kim, J. Won, and Y.S. Kang, "Olefin-induced Dissolution of Silver Salts Physically
Dispersed in Inert Polymers and their Application to Olefin/Paraffin Separation," Journal
of Membrane Science, 241, 403-07 (2004).
[27] H.W. Quinn, "Hydrocarbon Separations with Silver(I) Systems," in Progress in

Separation and Purification, E. S. Perry and C. J. V. Oss, Eds., 4, Wiley-Interscience,
New York (1971).
78 311 Final report

DE-FC36-04GO14151
[28] B.S. Friedman and R.F. Stedman, "Process for the Separation of an Unsaturated
Hydrocarbon from a Hydrocarbon Mixture," U.S. Patent 2,391,404 (December 25, 1945).
[29] G.F. Horsley, "Process for Separating and Recovering Olefines from Gases Containing
Same," U.S. Patent 1,875,924 (September 6, 1932).
[30] H.G. Van Raay and U. Schwenk, "Process for Separating Olefinic Hydrocarbons Using
Silver Fluoborate and Silver Fluosilicate Solutions," U.S. Patent 2,913,505 (November
17, 1959).
[31] A.E. Marcinkowsky, G.E. Keller, and S.K. Verma, "Olefin Separation Process," U.S.
Patent 4,174,353 (November 13, 1979).
[32] E.F. Steigelmann, "Membrane Process and Product," U.S. Patent 4,014,665 (March 29,
1977).
[33] J.C. Davis, R.J. Valus, R. Eshraghi, and A.E. Velikoff, "Facilitated Transport Membrane
Hybrid Systems for Olefin Purification," Separation Science and Technology, 28, 463-76
(1993).
[34] M.R. DuBois, R.D. Noble, and C.A. Koval, "Methods of Production of Novel
Molybdenum-Sulfide Dimers and Reactions of the Same," U.S. Patent 5,430,225 (July 4,
1995).
[35] R.B. Wilson, K. Meyer, and M.F. Asaro, "Unsaturated Hydrocarbon Separation and
Recovery Process," U.S. Patent 5,744,685 (April 28, 1998).
[36] M.A. Bennett, "Olefin and Acetylene Complexes of Transition Metals," Chemical
Reviews, 62 (6), 611-52 (1962).
[37] T.C. Merkel, Z. He, A. Morisato, and I. Pinnau, "Olefin/Paraffin Solubility in a Solid
Polymer Electrolyte Membrane," Chemical Communications, 1596-97 (2003).
[38] A. Morisato, Z. He, I. Pinnau, and T.C. Merkel, "Transport Properties of PA12-
PTMO/AgBF4 Solid Polymer Electrolyte Membranes for Olefin/Paraffin Separation,"
Desalination, 145, 347-51 (2002).
[39] T. Linnert, P. Mulvaney, A. Henglein, and H. Weller, "Long-Lived Nonmetallic Silver

Clusters in Aqueous Solution: Preparation and Photolysis," Journal of the American
Chemical Society, 112, 4657-64 (1990).
[40] S.I. Kantorowicz and R.M. Venner, "Metathesis: Refinery and Ethylene Plant
Applications," First Asian Petrochemicals Technology Conference, Taipei, Taiwan (May
22-23, 2001).
79 311 Final report

Separation of Olefin/Paraffin Mixtures With Carrier Facilitated Membranes

Uploaded by

Copyright:

Available Formats

Separation of Olefin/Paraffin Mixtures With Carrier Facilitated Membranes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Separation of Olefin/Paraffin Mixtures With Carrier Facilitated Membranes

Uploaded by

Copyright:

Available Formats

Separation of Olefin/Paraffin Mixtures with Carrier

Final report prepared by

U.S. Department of Energy

March 12, 2007

Award Number: DE-FC36-04GO14151

Project Title: Separation of Olefin/Paraffin Mixtures with Carrier Facilitated

Project Period: 11/01/2003 to 10/31/2006

Recipient Organization: Membrane Technology and Research, Inc.

Partners: SRI International ABB Lummus Global

Technical Contact: Tim Merkel

Business Contact: Elizabeth Weiss

DOE Project Officer: William Prymak

DOE Contract Specialist: Jean Siekerka

2 311 Final report

EXECUTIVE SUMMARY .............................................................................................................4

3 311 Final report

In addition to olefin conditioning, traditional carrier poisoning challenges must also be

4 311 Final report

5 311 Final report

In light of the importance of olefin/paraffin separations, and because of the costliness of

The spectacular separation performance of facilitated transport membranes has generated

6 311 Final report

7 311 Final report

8 311 Final report

9 311 Final report

Figure 1. Schematic representation of a facilitated transport membrane containing silver

10 311 Final report

AgBF4 concentration (vol%)

Figure 2. Mixed-gas ethylene/ethane selectivity in Pebax® 2533 as a function of AgBF4

2Ag + + H 2 ⎯⎯→ 2Ag 0 + 2H + Hydrogen reduction

In addition, silver salts can be reduced by light:

11 311 Final report

3.1 Preparation of the Polymer Electrolyte Composite Membranes

Figure 3. Schematic drawing and electron micrograph of a solid-state facilitated transport

12 311 Final report

Figure 4. Schematic diagram of the MTR casting machine.

Figure 5. Schematic diagram of MTR’s thin-film coating machine.

13 311 Final report

3.2 Membrane Stamp Permeation Measurements

Figure 6. Flow diagram of the mixed-gas permeation test system.

14 311 Final report

4. INITIAL PERMEATION AND SORPTION PROPERTIES OF SILVER SALT

AgBF4 Content Pure-Gas Permeances (gpu)

15 311 Final report

AgBF4 content (wt%)

Figure 7. Pure-gas permeances in Pebax® 2533+AgBF4 composite membranes as a function

16 311 Final report

AgBF4 content (wt%)

Figure 8. Pure-gas olefin and paraffin permeances in Pebax® 2533+AgBF4 composite

17 311 Final report

Gas permeance (gpu)

AgBF4 content (wt%) AgBF4 content (wt%)

18 311 Final report

AgBF4 content (wt%)

Figure 10. Pure- and mixed-gas ethylene/ethane selectivities in Pebax® 2533+AgBF4

The Figure 10 results, showing an increase in mixed-gas ethylene/ethane selectivities with

19 311 Final report

AgBF4 content (wt%)

20 311 Final report